Three completely different failures get called “cloudy solar lights,” and only two of them are worth your Saturday. A yellowed lens is a skin about 25 micrometres deep and comes off with wet-sanding. A crazed panel is a field of micro-cracks and needs filling, not sanding. A bulk-yellowed panel is amber all the way through the resin, and nothing you own will fix it. Spend two minutes telling them apart before you touch anything.
Two minutes of diagnosis
Take the light apart far enough to separate the clear dome or globe from the small solar panel on top, then run a wet finger across each surface and watch what happens as it dries.
| What you see | Wet-finger result | What it is | Verdict |
|---|---|---|---|
| White chalky film, mostly near the rim or where water sat | Scrubs off with vinegar | Hard-water scale | Easiest win. Rule it out first. |
| Clear dome or globe has gone amber; the plastic is still glossy | Wetting changes nothing | Photo-yellowed lens, surface layer only | Sand, polish, reseal. |
| Panel looks milky or frosted, colour still bluish underneath | Haze vanishes while wet, returns as it dries | Crazed surface resin | Fill it with a clear coat. Do not sand. |
| Panel is amber or brown through its whole depth | Wetting changes nothing | Bulk-yellowed encapsulant | Replace the head. Nothing works. |
The wet-finger test works because water has a refractive index close enough to the cracked resin to cancel the scattering. Fill the cracks and the haze stops. If water makes no difference, the light is being absorbed rather than scattered, and absorption spread through a volume cannot be polished off a surface. That single test decides most of what follows, and it is covered from the panel’s side in our guide to making solar garden lights brighter.
Which plastic am I actually holding?
This matters more than the method, because the same bottle that is harmless on one of these will wreck another.
| Part | Usually made from | How it fails outdoors |
|---|---|---|
| Clear dome, globe, lens | Polycarbonate (PC), acrylic (PMMA), or styrenics on the cheapest lights | PC yellows. PMMA essentially does not. Styrene yellows fast and goes brittle. |
| Body, collar, cap, stake | ABS, polypropylene, PC/ABS blend | ABS chalks and yellows; usually cosmetic |
| Clear film over the panel | PET or a thin fluoropolymer, on better lights | Delaminates or clouds at the edges |
| The panel itself | Poured epoxy potting over the cells | Crazes, then yellows in bulk. Standard epoxies are not UV-stable. |
Note what that table implies: a yellowed clear part is almost never acrylic. Röhm warrants PLEXIGLAS against yellowing for 30 years on the strength of PMMA being UV-stable as a molecule rather than because of a coating. So if your dome has gone amber, you are looking at polycarbonate or a styrene plastic, and polycarbonate is the one with a long list of chemical enemies.
Four tests, no lab
- Flex it. Polycarbonate bends and springs back. Acrylic is stiff and will crack or chip rather than flex. Styrene cracks with almost no warning.
- Tap it. Acrylic rings with a high, glassy tick. Polycarbonate gives a dull tock.
- Scratch a hidden spot with a coin. Uncoated polycarbonate marks easily. Acrylic resists it.
- One drop of 70% isopropyl alcohol on a hidden, slightly flexed edge. Covestro’s Makrolon data lists pure isopropanol as resistant on polycarbonate; acrylic charts rate isopropyl alcohol as only marginally compatible and it will bloom or craze under stress. Wipe it off after 60 seconds either way.
Restoring a yellowed polycarbonate lens
Here is the number nobody publishes: the discoloured layer is about 25 micrometres deep, because that is roughly how far UV penetrates polycarbonate before it is absorbed. The Madison Group, a plastics failure-analysis lab, puts it plainly — the discoloration “is principally a surface phenomenon, approximately 25 micrometers deep.” That is 0.001 inch. A quarter of the thickness of a sheet of copy paper. A third of a human hair. You are not grinding a lens down; you are skimming a film off it.
Which is why grit choice is simple once you see it against that number:
| Paper | Average abrasive particle | Use |
|---|---|---|
| P800 | ~22 µm | Start here only if the yellow is deep and even |
| P1000 | ~18 µm | Normal starting point |
| P1500 | ~13 µm | Second pass |
| P2000 | ~10 µm | Third pass |
| P2500 | ~8 µm | Final pass before polish |
Method, and it is the same one automotive headlight kits use because headlight lenses are the same plastic with the same problem:
- Wash with dish soap and lukewarm water. Dry. Never rub a dusty lens dry — you are just lapping grit into it.
- Tape off anything you do not want cut, especially the solar panel.
- Wet-sand with P1000, keeping the paper and the part flooded, in one direction only. Stop the moment the yellow is gone — it will look uniformly frosted, not amber.
- Repeat with P1500, then P2000, then P2500, each time at 90° to the last pass so you can see when the previous scratches are gone.
- Polish with a mild automotive polish. POLYVANTIS, which makes LEXAN sheet, explicitly names “a mild automobile polish” for minimising scratches and abrasions on polycarbonate. Covestro’s sheet manual adds one condition: alkali-free polishing paste.
- Seal it. See below. This is not optional.
The step that decides whether it lasts
Consumer Reports has tested headlight restoration kits repeatedly and the finding is consistent: they work, and the haze “will return, typically becoming noticeable after a year.” That is not a failure of the sanding. It is what happens when you remove the UV protection along with the yellow.
On a car headlight, what you sand off is a factory hard coat. On a $12 garden light, there usually is no hard coat — a moulded globe relies on UV absorber blended into the resin, and that absorber is concentrated in and consumed from the outer skin. Either way the conclusion is identical and, on a garden light, slightly more urgent: underneath the yellow layer is bare, unprotected plastic. Seal it with a clear coat that is explicitly sold as UV-resistant — a two-component automotive clear, or an aerosol clear labelled UV-resistant and non-yellowing. Two light coats, not one heavy one, and spray a test patch on the underside first, because some aerosol clears carry solvents that polycarbonate does not tolerate.
A crazed panel: fill, do not sand
If the haze disappeared under your wet finger, you have micro-cracks in the clear resin over the cells, and you fix them by refractive-index matching — flooding the cracks with something optically close to the original resin so light stops scattering back out. A thin coat of water-clear UV-stable acrylic lacquer or clear spray varnish does this.
Do not wet-sand a panel. The hazed layer on a potted mini-module is often only a few thousandths of an inch thick and the silicon is directly underneath. There is material to spare on a lens; there is none here. And do not reach for clear nail polish, which is the fix the forums recommend — see below for why it is the right idea in the wrong bottle. The panel-side cleaning routine is in how to clean the solar panel on garden lights.
A bulk-yellowed panel: stop
Amber all the way through, water changes nothing, and no coating helps, because the absorption is distributed through the resin rather than sitting on top of it. Adding a clear coat over it makes things marginally worse — one more interface to reflect at, and no light recovered. The cells underneath are usually fine; the window over them is not. Replace the head, or harvest the battery and LED. We cover what is and is not worth saving in can I repair a broken solar panel and how long solar lights last.
Every popular remedy, tested against the chemistry
| Remedy | Verdict | Safe on PC / PMMA / ABS? |
|---|---|---|
| Toothpaste | Works, but only on the top few micrometres. Non-whitening paste runs around RDA 70; whitening paste reaches 200 (the FDA’s suggested ceiling). Even at 200 it is a polish, not a cutting abrasive, so it will brighten a lightly hazed lens and will not touch a 25 µm yellow layer. | Yes / yes / yes — will dull high gloss |
| Baking soda | Mohs 2.5 and water-soluble, so the slurry loses its cut almost immediately. Dulls gloss without removing meaningful depth. | Yes / yes / yes |
| Magic eraser (melamine foam) | A micro-abrasive, not a chemical. Good on grime and scuffs; it will frost a clear lens and must never touch a panel. | Grime only, never on clear parts |
| Heat gun | No. Flame-polishing works on acrylic edges; polycarbonate blisters because of absorbed moisture, and a moulded globe distorts long before its surface flows. Heat also does nothing to the chromophores — the yellow is a chemical change, not a texture. | No / risky / no |
| Brasso | No, on polycarbonate. Its safety data sheets describe an ammonia odour, pH 11, and petroleum distillates with silica — an alkali, an amine source and an aromatic-bearing solvent in one bottle, which is three of the four things polycarbonate is documented to be damaged by. | No / acceptable / acceptable |
| Headlight restoration kit | The right tool. Same plastic, same failure, and the good kits include the sealant. Budget one per year unless you add your own UV clear coat. | Yes / yes / n/a |
| Clear nail polish | No. Two separate problems: it is nitrocellulose, which patent literature describes as having “a high potential for yellowing,” and it is dissolved in ethyl and butyl acetate — esters, which Covestro lists among the chemical classes that damage polycarbonate, and which acrylic charts rate “not recommended” at any temperature. The refractive-index idea is sound; this is the wrong material to do it with. | No / no / marginal |
| Acetone / nail polish remover | Hard no, and it is being recommended on page one of Google. Acetone dissolves acrylic and ABS on contact and swells polycarbonate. POLYVANTIS names acetone specifically in its “never use” list for LEXAN. | No / no / no |
| Hydrogen peroxide (the “Retrobright” trick) | An ABS technique borrowed from retro-computing, and a poor one: long-term follow-ups report the yellow returns, plus streaking. Harmless on a clear lens — polycarbonate is notably stable against 30% peroxide — and useless there, because it bleaches ABS rather than removing a layer. | Yes / yes / temporary at best |
| WD-40 | No. No abrasive and no chemistry for yellowing. It makes plastic look wet for a week, then holds dust. On a panel it is a light-attenuating film that collects grime. | Pointless on all three |
| DEET bug spray | No. The NPIC technical fact sheet states flatly that “DEET is a plasticizer and can damage certain rubber, plastic, vinyl, or elastic materials.” It softens the surface, which looks like restoration for about a day. | No / no / no |
| White vinegar | Yes, for what it is for: hard-water scale, which is a genuine cause of “cloudy” in most of Southern California. Polycarbonate is rated resistant to 10% acetic acid. It will not touch yellowing. | Yes / yes / yes |
| Ammonia glass cleaner | No on polycarbonate, fine on acrylic. See below. | No / yes / yes |
Why ammonia is the one that actually matters
Ammonia does not stress-crack polycarbonate. It degrades it. Polycarbonate is an ester of bisphenol A and carbonic acid, and amines attack that ester linkage directly by transamination — aminolysis — chopping the polymer chains whether or not the part is under load. Covestro’s data rates 0.1% ammonia in water as non-resistant, and the Makrolon technical manual lists “ammonia and amines, as well as their solutions” among the things that damage polycarbonate outright.
That is why this matters more than it sounds: a stress crack needs stress, so you can sometimes get away with it. Aminolysis needs nothing. And it is entirely plastic-specific — acrylic charts rate ammonia and 5% ammonium hydroxide as good to excellent. The same spray bottle is fine on one clear dome and destroys the next one. Our solar lights cleaning guide lists the rest of what should never touch either part.
Keeping it clear
- Reseal after every restoration. Unsealed, you are back in a year. This is the single difference between a repair and a chore.
- Wash with soap and lukewarm water, not dry. Never rub a dry, dusty lens, and never clean it in direct sun — POLYVANTIS warns that cleaning polycarbonate hot causes staining.
- Move the worst-exposed fixtures. A light in all-day western sun in Garden Grove will yellow years before one under a eucalyptus. Rotating pathway lights around a run once a season evens the wear.
- Buy the lens, not the lumens. A moulded globe with UV absorber blended in is a consumable. Glass or acrylic optics are not. That distinction predicts lifespan better than any brightness claim on the box, which is why spot lights with glass lenses hold up so much better than decorative plastic ones.
- If it went dim rather than yellow, this was the wrong article. Start with solar lights not working; the usual answer is the battery. And if the fixture itself is sound but the stake snapped, see making replacement stakes.
Restoring a lens is genuinely satisfying, and it takes about fifteen minutes per light once you have the grits laid out. Just do the diagnosis first. Half the lights people sand should have been sealed, and a quarter of them should have been thrown away.
Sources
- The Madison Group - Yellowing of Polycarbonate madisongroup.com The 25-micrometre figure, and that photo-yellowing is a surface phenomenon limited by how far UV penetrates
- Exolon Group / Covestro - Makrolon Sheet Technical Manual exolongroup.com What polycarbonate resists against what damages it - alkali, ammonia and amines, aromatic and halogenated solvents, aldehydes, ketones and esters - plus the alkali-free polishing paste requirement
- POLYVANTIS - LEXAN Sheet Cleaning Instructions 2025 polyvantis.com Never use abrasive or highly alkaline cleaners, never use aromatic or halogenated solvents or acetone, mild automobile polish endorsed for scratches, and do not clean in direct sunlight
- Consumer Reports - Do car headlight restoration kits really work? consumerreports.org That the kits work but the haze returns, typically becoming noticeable after a year, because the protective layer goes with the yellow
- ePlastics - Chemical resistance of Plexiglass acrylic eplastics.com The acrylic ratings behind the contrast table: ammonia good, 5% ammonium hydroxide excellent, hydrogen peroxide excellent, isopropyl alcohol fair, acetone and ethyl acetate not recommended
- Rohm - PLEXIGLAS UV resistance plexiglas.de That PMMA is UV-stable through the whole sheet rather than by coating and carries a 30-year warranty against yellowing - the basis for 'a yellowed clear part is almost never acrylic'
- NPIC - DEET technical fact sheet npic.orst.edu That DEET is a plasticizer and can damage certain rubber, plastic, vinyl or elastic materials
- Ted Pella - Abrasive grit grading systems tedpella.com The FEPA P-grade average particle sizes in the grit table, from P800 at about 22 micrometres to P2500 at about 8
Questions people ask
Can yellowed plastic be whitened?
On a clear lens, yes, but you remove it rather than bleach it. Polycarbonate yellowing sits in roughly the top 25 micrometres, so wet-sanding through P1000 to P2500 and polishing takes it off completely. On opaque ABS housings, hydrogen peroxide bleaching does lighten the colour, but long-term follow-ups show the yellow returns along with streaking. On a panel that is amber through its full depth, nothing works.
Does WD-40 clean solar lights?
No. It has no abrasive and no chemistry that addresses yellowing. It leaves an oily film that makes plastic look temporarily wet and clear, then collects dust and pollen for months. On a solar panel that film measurably cuts the light reaching the cells, so it makes charging worse rather than better. Use dish soap and lukewarm water, and a restoration kit if the lens is actually yellowed.
How can I reverse the yellowing of plastic?
You cannot reverse it chemically - the yellowing is broken polymer producing compounds that absorb blue light, and those cannot be un-broken. What you can do is remove the layer they live in. That only works when the degradation is confined to the surface, which is true of a clear polycarbonate lens and false of an epoxy-potted solar panel. Test by wetting it: if water changes nothing and the part is amber throughout, replace it.
Does toothpaste work on cloudy solar lights?
Partly, and it is safe on polycarbonate, acrylic and ABS alike. Non-whitening toothpaste runs about RDA 70 and whitening paste up to 200, which makes it a polish rather than a cutting abrasive. It will brighten a lightly hazed or scuffed lens and it will not remove a 25-micrometre yellow layer no matter how long you rub. Use it as the final step after sanding, not instead of sanding.
Will vinegar or baking soda remove the yellow?
No, but vinegar fixes a different problem that looks identical. White chalky film near the rim of a light is hard-water scale, and dilute acetic acid dissolves it without harming polycarbonate, acrylic or ABS. Rule that out first, because it takes two minutes. Baking soda is Mohs 2.5 and dissolves as you work, so it dulls the gloss without cutting deep enough to matter.
Is it safe to use ammonia-based glass cleaner on solar lights?
Only if the clear part is acrylic. Ammonia attacks polycarbonate by aminolysis, chemically splitting the polymer chains, and it does so whether or not the part is under stress - Covestro rates 0.1% ammonia in water as non-resistant on Makrolon. Acrylic, by contrast, tolerates ammonia well. Since a yellowed clear dome is almost always polycarbonate rather than acrylic, assume the answer is no.
Is restoring worth it, or should I just replace the light?
Restore the lens; replace the panel. A yellowed globe takes about fifteen minutes and costs the price of sandpaper and a can of UV clear, and the result holds for years if you seal it. A bulk-yellowed panel is a new light - the encapsulant is the part that failed and it cannot be recoated. If the light is dim but nothing is yellow, the battery is the usual culprit and that is a cheaper fix than either.
