Lemi Shine is magical. We must have the hardest water ever. Any harder and I’d be able to walk on it. Our dishes had the grossest white film on it that just kept getting thicker and grittier and grosser. I thought we were going to have to wash dishes by hand or buy new dishes. Then we tried adding Lemi Shine. No joke, after 2 or 3 cycles, the dishes look like new. I wish I had before and after pictures. Imagine an opaque drinking glass next to a crystal clear drinking glass. Actually, the picture on the bottle could have been taken in our kitchen.
So I wanted to know what voodoo powder is in Lemi Shine that allows for such magical transformations to happen. Looking on the innertubes, people making home-made dishwasing detergent often have a line like ‘add Lemi Shine or lemon Kool-Aid packets.’ So that’s weird. The label on the bottle says it contains ‘natural fruit acids and citric oils.’ Ok, so maybe the lemon Kool-Aid isn’t so weird. Here’s a list of natural fruit acids.
So I turned to the MSDS in hopes that it would divulge the ingredient list. The Lemi Shine MSDS was really easy to find, but the composition section reads:
Well, thanks. That could not be less informative. So my bottle contains between 61-105% of something? (maybe the rest is chemical free. oooooh!) The only actual quantitative information the MSDS provides is a pH of 3 and some LD50 data: Compound1: 3000 mg/kg (rat, oral); Compound2: 2840 mg/kg (rat, oral), 5000 mg/kg (rabbit, dermal – what did that experiment look like?)
So I guess since I’m a chemist, I should bring some into lab and figure it out myself, eh? So I did.
I asked around on Twitter what spec (practically, not ideally) would people use if they had an unknown colorless solid and wanted to figure out it’s composition. Got lots of great suggestions and promptly put them to use. I also posted my results in real time on Twitter so everyone could play along (#WhatsInLemiShine). I started with appearance. It’s a colorless solid with a strong lemon odor. The free-flowing granules are rather large, imho. Having used several bottles in my dishwasher already, I know it’s not very hygroscopic. No, I did not taste the Lemi Shine (though I’m sure it’d be quite harmless)
Then I ran solubility tests. You’ll be relieved to know it’s soluble in water. Here’s the rest of the solubility data:
@BranVanChemist asked what the pH was. The pH of a solution of 26mg of Lemi Shine in 1mL water is approximately 2:
About this time, @DrBodwin suggested I run a GC/MS on the filtrate from the dichloromethane solubility test. The entire granule may not have dissolved, but maybe some trace organics did. Also about this time, I noticed the ethyl acetate getting cloudy just above the granules. So I prepped GC/MS samples of the Lemi Shine completely dissolved in methanol, and the filtrate from the dichloromethane and ethyl acetate solubility tests.
While those were running (I <3 auto-samplers), I ran an IR and took a mp. The Lemi Shine melted at 150-153 degC. Here’s the IR:
Not the most inspiring IR, but reminds me of the ‘hairy beard’ effect a la Jon Chui’s pictorial guide to IR. Then the GC/MS results started coming in. The dichloromethane GC/MS showed no peaks, so really nothing must have been soluble in dichloromethane. The methanol GC/MS was inconclusive. There were several low-concentration components, but nothing that screamed major component. The ethyl acetate GC/MS also showed a few peaks; however, not the same peaks as from the methanol GC/MS. My guess is this is the citric oils, probably for fragrance? Though I haven’t yet conclusively identified any of the components of this GC/MS. A note about the ethyl acetate GC/MS. I can only include 4 MS on the print out. Peak 1A and the unlabeled peak to its right have the same MS. Similarly, peak 2A and the unlabeled peak to its right have the same MS. Methanol on left, ethyl acetate on right:
I had a guess going into the GC/MS based on online discussions, but my guess was not confirmed through GC/MS. Sources (read: wikipedia) said my guess decomposes above 175 degC, so that may explain why it didn’t show up on GC/MS. An LC/MS would perhaps give me more information, but we don’t have a walk-up LC/MS.
Since it wasn’t soluble in CDCl3, but it was soluble in DMSO, I next went for HNMR. I was warned the NMR could be quite complex depending on how many components there are in Lemi Shine. Surprisingly, the NMR was quite clean. So if there is a mixture of components, this tells me there is one major component. I did also take an NMR of the filtrate from the CDCl3 solubility test. Just some water in the spectrum, no other components. Like dichloromethane, Lemi Shine really is completely insoluble in CDCl3. I would have expected some of the citrus oils to be soluble in CDCl3, though, so that’s perhaps a bit unusual. DMSO on left, CDCl3 on right:
@stuartcantrill noted that the residual proteosolvent peak for DMSO (which appears around 2.5 ppm) should have 5 lines, not four. But I think the apparent quartet is a real signal from the Lemi Shine, not the solvent peak. They just happen to appear in the same area and thus overlap. It really is noteworthy how clean the DMSO NMR is. Chemists often lament the trace residual ethyl acetate or dichloromethane contaminating their NMR spectra during synthesis. The flat baseline indicates Lemi Shine has one major component and if there’s anything else, it really must be closer to 1%.
@JessTheChemist asked for a CNMR, which I was able to obtain in only ~20 min on our not-very-strong NMR. The baseline’s noisy, but again, this indicates only one major component. This time, the DMSO solvent peak is quite visible:
The HNMR and CNMR confirmed my initial guess as to the identity of the major component of Lemi Shine. I’ll summarize the data here, then I’ll hide my guess below if you want to comb through the data and decide for yourself first.
|appearance||large, colorless granules with lemon odor|
|pH||~2 (26 mg in 1 mL water)|
|melting point||150-153 degC|
Here’s my guess based on the compiled data. It’s in white font so highlight it to read it: citric acid
***Here be spoilers***
Don’t read on if you still want to guess yourself
My guess is based on many online diy dishwashing detergent sites saying you can buy Lemi Shine or straight citric acid. Also, Sigma-Aldrich’s entry for citric acid lists a mp of 153-159 degC, and has a reference NMR. The NMR matches perfectly.
I posted all this to twitter, and there were high-fives and champagne all around. Now, I should have known this, but Lemi Shine has its own Twitter handle (@Lemishine) and they responded shortly after I posted my conclusion:
My first instinct was: they’re lying. And I told them so (nicely). But it bugged me for the rest of the night. If it’s really not citric acid, what is it?
A citrate salt like sodium citrate might still make sense. Could explain GC/MS data. But would it give pH ~2? Certainly not sodium citrate tribasic. But maybe sodium citrate monobasic could turn my pH paper red. But CNMR would look the same, too. But integration in HNMR would be off. There’d only be a 2:1 ratio of acid:alcohol protons. Not that broad singlet integration is an exact measurement, but I’d expect it to be a bit closer to 2:1 not 3:1.
Maybe they were just being cute. In 140 characters, I only said it was citric acid. There are probably some trace fragrant oils or something to give the 1-5%, so perhaps I was wrong because I wasn’t technically right?
Someone on Twitter suggested polycitric acid (though I can’t find the original tweet anymore). That could certainly make a lot of sense. Only spectra I’d expect to be different would be peak heights in HNMR and/or CNMR. Some protons are going to be lost in polymerization, and if it was polymerized through ester bonds, there’d presumably be some linker with protons and/or carbons of it’s own, yes?
So where does that leave us? What’s in Lemi Shine? I’m sticking with citric acid. If someone sees something else in my data, I’d love to know. On Twitter, we decided there need to be more live-Tweeting of chemistry and made the hashtag #RealTimeChem.
@JessTheChemist is chronicling the real time chem stories via storify, so check it out there, too. Today @sciencegeist is live-Tweeting his synthesis of the non-natural amino acid L-azidohomoalanine
Thank you all for your comments and follow up tests I should run. I ran most of them! Here are my results:
Brandon suggests a ninhydrin test for the ammonium cation to see if Lemi Shine is an ammonium salt. Ninhydrin is one of the chemicals used in forensic development of fingerprints and reacts with nitrogen to create a red/purple color. So I spotted Lemi Shine on a TLC plate and dipped it into a ninhydrin stain. Even after heating, the TLC plate showed no evidence of nitrogen.
Stewie Griffin also suggested a test for the ammonium cation. He suggested the USP test. Since I teach at an R7 school, I was quite surprised we actually had all the reagents necessary for the USP test. So I did as Stewie suggested and bubbled through the indicator solution – didn’t turn yellow. I also added the cobalt reagent – didn’t give yellow precipitate. I’m confident with these two tests to say Lemi Shine is not an ammonium salt.
Trent Wallis suggested a quick, even microscale, titration for more accurate data. If I’m going to get some data, I’m going to try to get it right, so a ‘quick and dirty’ titration isn’t in my bones So I reached waaaaay back to remember how to do titrations. Something about a buret. Make sure the stopcock’s closed. Oh yeah, I guess I should standardize the NaOH. While I didn’t standardize in triplicate, I calculated my solution of NaOH to be 0.0942 M.
Then I titrated 0.2278 g Lemi Shine in 30 mL H2O with this 0.0942 M NaOH. After every 0.1 mL, I measured the pH and recorded it in Excel. After 19 mL, I realized this was very very tedious, so I measured the pH every 0.2 mL. At 33 mL NaOH, the pH started rising more rapidly, indicating I was approaching the endpoint, so I went back to measuring the pH every 0.1 mL. The endpoint came around 36.1 mL 0.0942 M NaOH. I finished out the 50 mL buret for completeness’s sake. Here’s my titration curve. Source data available upon request.
I found a titration curve someone else put together for citric acid. Note the similarities:
The exact pH of the Lemi Shine (0.2278 g in 30 mL H2O) before titration was 2.3. If we take the equivalence point to be 36.1 mL (0.0034 mol NaOH @ 0.0942 M), then the equivalent weight of the acid is (0.2278 g/0.0034 mol) = 66.99 g/mol. Assuming it’s a triprotic acid (based only on the titration data, an invalid assumption), that would give the molecular weight of the acid to be 200.97 g/mol. The molecular weight of citric acid is 192.124 g/mol. More on that discrepancy later.
Mitch said I should just taste it and be done. No.
TSvn suggested adding Lemi Shine to acetic anhydride in pyridine. If it’s citric acid, the solution should turn red. So I did. And it did! Further evidence for some sort of citric acid?
Shakeel suggested a flame test. Cool! Unfortunately, Lemi Shine doesn’t change the color of the flame. On one trial the flame flickered green a bit, but I’ll chalk that one trial up to the copper wire I was using. I also ran a flame test on a standard solution of citric acid from the stockroom and on sodium citrate tribasic from the stockroom. The standard citric acid also did not change the flame color – except for one trial which flickered green like before. The sodium citrate, though, turned bright yellow as expected. Potassium should turn violet, lithium should turn red, magnesium would be white, and calcium would be a red/orange (wikipedia)
Finally, fledarmus suggested getting an NMR of Lemi Shine, an NMR of a citric acid standard, then mix the NMR tubes together and see if the peaks match. So I did. And they do! The citric acid standard was the monohydrate, so the HDO peak shows up in the monohydrate NMR
Citric Acid Monohydrate
Lemi Shine + Citric Acid Monohydrate
Note how the AB system for the diastereotopic methylene protons completely overlap, with no shoulders or splitting, etc. I’m convinced it’s citric acid, not some salt of citric acid.
Now, about that pesky titration data. Let’s assume it’s citric acid. The molarity of 0.2278 g citric acid in 30 mL would be 0.03952 M. Different websites have different pKa values for citric acid, an average of 7 of them gives a pKa of 3.122. That’s a dissociation constant for citric acid of 0.0007553. Thus the pH of a 0.03952 M solution of citric acid would be… 2.3, exactly the pH I noted for my sample.
But if the endpoint is at 36.1 mL of 0.0942 M NaOH, that gives an equivalent weight of the acid to be 66.99, which would correspond to the molecular weight of a triprotic acid of 200.963 g/mol. This is 8.839 g/mol greater than citric acid (192.124 g/mol).
If we assume citric acid, we essentially have zero unknowns in our equations, so we can check the values against each other by asking ourselves GenChem questions. How many mL of 0.0942 M NaOH are required to completely react with 0.2278 g citric acid? With a 3:1 stoichiometry, the equivalence point should have occurred at 37.8 mL. That’s a difference of 1.7 mL – that’s a lot of mL.
So, if I’m pretty sure it’s not a citrate salt (the pH wouldn’t have been 2.3 for a citrate salt anyway), maybe it’s a hydrate of citric acid? If that were the case, then I wouldn’t really be titrating 0.2278 g citric aicd. For the monohydrate (MW = 210.14 g/mol), the mass fraction of citric acid is 91.43%. So a 0.2278 g sample of the monohydrate would really be 0.2083 g citric acid. How many mL NaOH would I need then? 34.5. Now I’m off by 1.6 mL in the other direction.
My endpoint was between the anhydrous and the monohydrate… what if it’s the hemihydrate? The mass fraction of the hemihydrate (MW = 201.133 g/mol) would be 95.52%, so a 0.2278 g sample would be 0.2176 g citric acid. How many mL NaOH would I need then? Why 36.07 mL! If the equivalent weight was 66.99, and the MW would be 200.96 g/mol, that’s 0.2 g/mol off the MW of the hemihydrate (201.133 g/mol).
So! My ultimate final guess is citric acid hemihydrate. I think I’m done testing Lemi Shine. So if someone else wants to run other tests, be my guess. Just be sure to post your results here so we can know!