etching/frosting LEDs?

I have a bunch of water-clear LEDs that I'd like to uniformly frost for diffuse illumination. Has anybody any experience in chemically frosting them?

'Appearance counts', so sitting down and buffing them with sandpaper is rather out of the question. I haven't purchased anything yet (save the LEDs) and am looking for recommendations for a good etchant.


LEDs are not so expensive. Why not buy diffuse LEDs in the first place?

It might not help but I haven't found a solvent to touch them yet. Sand paper however is a good alternative, you don't notice any scratches when they are on. The trick it to use a very fine grade sandpaper.

LEDs are not so expensive. Why not buy diffuse LEDs in the first place?

I should have mentioned they're 15 candela warm-white ones, this is a one-off project, I already have a bazillion of them, and this is an experiment to see if they'll better-suit a particular application if they're diffuse - so I can't see the sense to trying to look for some.

I've tried sulfuric (80%), nitric (70%), acetic (glacial), phosphoric (80%), HCl (33%), even hydrogen peroxide (25%) with zero results - it's been a busy day. For the time being, until I can discover anything that WILL etch the epoxy (and I don't have access to a sandblaster), I'll just be buffing them one-by-one as randomly as possible with fine Scotchbrite™ - something possible with a couple-dozen pieces but not desirable if I ever need a greater quantity (where I could 'batch' them with a simple dip/rinse/dry/repeat in an etchant).

What about something like this?:

Alternatively, instead of sanding them with sandpaper, what about using a buffing wheel (figuring out how to mount/steady them for the work might take some creativity; perhaps clamping the leads of the LEDs between a couple of pieces of flat bar stock would work?):

...loaded with some compound (maybe 120 grit or a bit higher?):

/note: I used Rustoleum/Caswell as example sites to show what I mean, I have no relationship or anything with either...

I’ve tried sulfuric (80%), nitric (70%), acetic (glacial), phosphoric (80%), HCl (33%), even hydrogen peroxide (25%) with zero results

You need an organic solvent. I have tried acetone, carbon tet, and a few other brand stuff without finding one.

I would suggest using a sand blaster to etch them. If you poke a bunch of them into a styrofoam sheet you should be able to whip through them pretty quickly.

As far as organic solvents you could try I would suggest MEK (PVC cleaner) if you have it available. It'll still probably take a lot of soaking to etch them.

superglue will fog them up ... maybe you can dip them as the fumes are not reliable

All done! I went the Scotchbrite™ route since I have to get this thing finished. NOT as difficult as I thought it would be (tedium: yeah, a little).

I picked up a sheet (8"x10"?) of their maroon-colored 'super fine' material (not soap-impregnated like a kitchen scrubbing-pad), pinched each LED between thumb and forefinger, and lightly traced a circle (about 4" diameter) while slowly 'orbiting' my wrist to keep the epoxy from developing a concentric pattern of scratches. I slowly orbited it from about 60º to vertical, and back, in about 20 'circles' (kind of reminded me of my old telescope mirror-grinding days). Pressed into the pad not even 1mm - a very light touch.

This removed quite a lot of the epoxy, so I dulled-down the pad by working it for a while with a cutoff piece of the pad to remove some of its 'bite' (definitely can feel a big difference between the two sides now). This worked better - the same grinding cadence gave me an evenly-frosted surface extending down the side of the package to almost the level of the 'cup' containing the chip. Under a 10x magnifier the very fine surface scratches appear to be well-randomized and I see no 'hot spots' on the surface of the epoxy.

I lit up a clear one in series with a frosted one (i.e. same current): the clear one has maybe a 15º dispersion (and is obviously dark off-axis, with several refractive/reflective zones) and the frosted one, while not quite as bright end-on, maintains very near even brilliance to more than 60º off-axis. In the future, I'll probably try to manufacture some kind of tube-shaped holder so I can buff an LED all the way down the side to the flange without sacrificing skin or fingernails. :astonished:

In all: I am well pleased for the total effort of maybe 1 minute per-LED on the dulled-down side of the pad. If anyone else tries Scotchbrite™ (or finds an etchant) please post your results here. Thanks to all for your comments.

the maroon pad is my best friend (green sux), and this is nearly the exact same way I defrosted a LED maybe last year, though I thought I would not want to do a batch of them

In case anyone comes looking another option is to go with a glass frosting spray like this. The only problem with this type of spray is that it scratches off fairly easily. So you may need to put a top coat of something like this to get a tougher surface. All depends on usage.

I went the Scotchbrite™ route

For those of us not in the U.S. any chance of a link so we can see what one of these is?

Google scotch-brite pad, they are made by 3M

For example

Thanks, we don't have that brand in the UK although we do have similar sorts of products now that I see one.

Actually, the stuff I used is their '7447' material: