making a reflective ring on a XM-L led-dome by silver deposition

I like to experiment, and was thinking of trying this for quite some time. A few weeks ago I did the actual experimentation, but the results were a bit disappointing so I hesitated doing the write-up. But I believe it is just fair to not just report your successes, but also the failures, and it involved quite some work, so here we go :-)

The idea was to deposit a silver layer directly onto a led-dome, to accomplish that the light going sideways (and is lost when the led is used in a aspheric lens flashlight) is reflected back to the die and so increase the illuminance. This has been succesfully done before with a ring shaped globular reflector outside the led, it is even patented, with a fancy name. Depositing the reflecting layer directly on the dome has been mentioned once on CPF a few years ago (I looked it up), but as far as what I can find, no one has actually tried it.

The chemical reaction that I used is the so-called Tollens reaction, used traditionally to make mirrors until somewhere in the 20th century, then they started making the now common aluminium mirrors that do not have that slight yellow colour. I demonstrate the reaction every now and then in the seconday school where I work as a technician, this is my first attempt of trying it on something that is not glass. Briefly: the substrate is cleaned with soap and then with 67% nitric acid, a silver nitrate solution is added, then a 25% ammonia solution, this gives a brown deposition, keep adding until clear , add a potassium hydroxide solution, brown stuff is deposited again, add again 25% ammonia until clear, then add a glucose solution and keep stirring until the silver layer is formed.

The first attempt was on a XM-L on a 20mm star. The result:

So to be usefull the top of the dome had to be cleaned. Further, the led had to be reflowed on another board, the silver gives shorts everywhere. I scraped the silver off th top of the dome with a tooth pick:

The cleared area of the dome was not crystal clear after that, but it was not too bad. I reflowed the led on a Sinkpad and tested it in series with an untreated XM-L to see any difference:

The treated led had a slightly warmer colour, but it was nothing obvious, and on the picture there is no visible difference leftover. It seemed that the treated led had a bit less illuminace than the untrated one, this was not promising for the method.

Second attempt:

A XM-L2 3C was used, and I wanted the clear part of the dome untreated, so this time I covered the top bit of the dome with a piece of 'parafilm' , which is thin parafin sheet. It adheres a bit to the dome without really sticking to it:

After the silver plating (the solder pads and lettering did not like the nitric acid treatment :evil: ):

That looks much better than the first attempt! Here it is compared to a untreated one:

But reflowed on Noctigons this time. The tints compared did again not look much different (led tester at 2mA):

I soldered the boards in a Cofly 18650 zoomie with a Qlite revA driver to test the numbers, first the untreated led board, then the treated one:

XM-L2 3B XM-L2 3B with silver ring attempt 2

current: 2.98A 2.97A

zoomed out: 586 lumen 533 lumen

zoomed in: 306 lumen 305 lumen

throw zoomed in: 13.1 klux at 1meter 14.4 klux at 1meter

So that is not all that impressive, is it? A slight improvement of throw, but nothing like doubling the number, what I hoped for of course :-(

It is impossible to see the inside of the silver layer from outside the led, so I dedomed the treated led to see how the reflective layer looked like from the underside of the dome. And it looked a bit golden, not good. Perhaps the silver deposition had been too fast, and some brown dirt had come into the layer.

Here's a stereo picture of the dome from the underside. (and it is a difficult one: the removed dome is cut in half and put under water to be able to see the layer, hence the air bubbles everywhere)

So I went a ahead with attempt nr. 3, I had the untreated XM-L2 3C leftover, it was already on a Noctigon, so this time I wrapped the entire board in parafilm, put a the piece on the top of the dome again, leaving only the area to be silver plated exposed. This time I forgot the nitric acid claeaning, but that did not seem to matter in the end. Usually I heat up the reaction mix a bit to help the speed, I did it now at room temperature, letting the silver layer deposit much more slowly.

Looking even better than the last time. but the numbers still disappoint: current 3A, zoomout: 585lm, zoom in: 330lm (= slightly increased), throw at 1meter: 15.7 klux. So the numbers are a bit better again, but still not impressive at all, compared to the untreated led, the throw went up by only 20%.

I had the dedomed led leftover from the second attempt, I measured that one as well: current 2.94A, zoom out 512lm, zoom in 266lm, throw at 1meter 25.4 klux. Now that is what I call a real improvement. The silver plated led did not do anything close to what dedoming does.

At this point I gave up. I can not see how I can do a fourth attempt that is any better than this. And I ran out of spare XM-L2's.

Still I got way further than I expected beforehand, I even was not sure if the silver would deposit at all on a silicon dome, and if the parafilm would prevent the silver plating on the covered bit, and if the 67% nitric acid would destroy the led (it did not, but also appeared not be be neccessary in the end). It was fun doing all this, but quite a disappointment that I did not succeed, especially not because I have many ideas why it failed but no idea what is the right one.

I hope you enjoyed reading about these wanderings outside the commonly done mods, even though it failed :-)

It’s all coming together now. djozz the mad technician. The kids at your school must love you. Thanks for the effort here in again an experiment not many here including me could achieve.

Thanks for reporting on this interesting project!

A few kids have me seen doing this, but they have no idea what this is all about, they find me a bit weird. Actually the really mad one is the chemistry teacher, I am usually the one preventing him from blowing up the school, we usually do a sort of role play together with him having the insane look in his eyes :-)

Only thing I might try is half-dedoming if possible. Like fully plating it and then cutting off the top half of the dome. Maybe that would get you some benefits of dedoming and adding the reflected light lumens? Just an idea, and then you would have to worry about covering the top of the dome either. You would need a good blade and a holder at the right height. Maybe one of those lumen-robbing, too tall centering rings.

Very interesting work, got me thinking, would some kind of high temperature white paint work?
If you held a tube in place, maybe a straw that was the right size to cover the top portion of the dome then carefully paint around it would be simpler and save trying to clean the top of the dome. It may also give a sharper line around the dome. Could this technique be implemented with your silver?

Thanks for sharing

I dont' think that is a failure at all. 20% increase in throw is awesome! I will bet that if you play around with how much of the dome is covered (aperture size), you will be able to improve that number quite a bit. In the end it may be easier to produce negatives of the dome and coat them instead. Now you have me thinking...

Such a cool idea and test. Thank you for sharing. I wish my successes were as good as your failures.

Why bother with silver deposition at first try? It’s like taking leaps for your first steps.

IMO you can get proof of concept by simply using chrome paint. Apply three layers or so chrome, then cut off a bit of the dome top. I’ve been meaning to do this but as all my other projects… hahaha… probably next year.

And secondly, proof of concepts also can be done using cheap cheap XPEs. Much easier on the wallet, you can buy bulk for experiments, they’re like 90cents each when buying lot of 10 (Aliexpress).

But I forget my manners. Thanks for doing this! It is also very nicely documented. Successes are built on constructive failures, mind you… so regardless of results this is a good foundation and contribution. Thanks again for doing this, and publishing it.

Great experiment! It wasn't a failure, you collected data which should help understanding why you didn't see the results you where hoping for. Thanks for the report.

On the last attempt, it seems to me that the clear area is way too large.

You can see practically the entire die. Ideally, I think you should only see the very center. In throw mode, the lens projects an image of the die. Bigger die, bigger image. If all the lens 'sees' is the center portion, say about the size of an XR-E, then the image size will match that of an XP-E. Additionally, a circular clear window on the dome should project a round image. What shape image did it throw? If it was square, the window was still large enough for the lens to 'see' the entire die. I think you need a much smaller window.

Thanks for the replies. You make me think it could be finetuned, perhaps I will try that at some later point. About some of the suggestions:

-white or metallic paint would work, but I guess a bit less well than a deposited mirror.

-in another project I had a go with dedoming by slicing very neatly with a spacer and a razor blade. The surface never comes out really clear and flat, not good for throw. And even then, it being flat instead of curved will not increase throw, as Dr. Jones explained to us in his lecture on throw.

-reducing the aperture will help, but my goal in this experiment was maintaining the die size ( or actually hotspot size) of the flashlight while making it brighter. I could reduce the aperture while improving center brightness of the hotspot up to the point where the center of the die can just still be 'seen' by the edge of the aspheric lens, in the used flashlight in zoom in modus that is a cone of 58 degrees, that would be an aperture that is still not that far from the used one here. But if spot size would not matter first thing to do is simply put in a xpg or xpe, at the 3A used their dies have a much higher illuminance. One fun experiment would be such a ring on a fully driven xpe2, with the smallest effective aperture, see how far I can stretch the meaning of 'pencil beam'

-because the aperture is quite a bit above the die, you still see a square image if the die is in focus, if you focus on the aperture, the hotspot is round but you already loose some throw because the die is out of focus.

Would it not be simpler to put a hat with a hole in it on the led? Or is the purpose purely to coat it with a substance?

The principle of this ring is not just block light, but to reflect light that is going sideways, and would be lost in an aspheric, back into the led die, some of that light will again cause fluorescence of the phosfor layer and so increase the brightness of the die.

The hat could be shiny on the inside to reflect back just as the silver coating would do. In fact I would assume you could coat it in silver as the OP is attempting to do. The hole being in the middle of the hat to allow light forward.

I get the sense Im missing something though.

Ah, I see, such a hat would work very well, in fact you just reinvented the Wavien Collar. It is out there and patented, I even believe that if I would make leds the way of the OP and try to sell them I could be drawn to court by them, because of violating their patent.

I now see that I didnt see what you see.

Thank you djozz for doing this very special and troublesome experiment. I always enjoy reading your experiments, like the crash-testing of various LEDs does give us very good information and reference. :slight_smile:

By the way maybe you should measure the lux reading of the same LED before and after you deposit it. Comparing a LED to another LED may be a little inaccurate as there might be some production variations among them.

“By seeking and blundering we learn.” ― Johann Wolfgang von Goethe

I agree with bibihang... And not jus little inaccurate... It can be more than 10% difference in brightness among same batch of same emitters so this is something to be considered in future experiments.

Great job Djozz... You are one of the guys that make this forum so interesting.

thank u for this very interesting post. I think it will be nice if u mirror again all led surface and just cut a small piece in front of it. than measure result, than cut again a lttle biger than measure again and so on. I believe it will be very informative. again thank u for this ingenious experiment.

hmm, I think this was quite a success. The fact that you consistently saw an increase suggests that what you’re doing is actually having an effect. You also increased throw without reducing lumens (downside of dedoming from what I’ve read), which is a pretty good definition of having your cake and eating it :slight_smile: