Skip to main content

Chemical News

Sodium Hydride in Aprotic Solvents: Look Out

Here’s a safety warning for my fellow synthetic organic chemists. It’s a reagent combination whose hazards have been noted before, but a lot of people don’t seem to know about it: sodium hydride in DMSO or other polar aprotic solvents.

And yeah, I’ve used that exact combination, too, many times. But I did those reactions (for the most part) before I was aware of the possible hazards, and they have mostly been on a rather small scale and mostly at room temperature. Scaling up any such preparation, especially with heating, is a very bad idea. Severe explosions have been reported for over fifty years, as the new paper linked above details, but people still use these conditions without knowing that. This paper notes that the major organic chemistry titles publish dozens of papers a year with these reagent combinations in the experimental sections.

When you use NaH/DMSO you’re forming the dimsyl anion (deprotonated DMSO), and that is not a stable species (as was first reported in the mid-1960s). If you know how to read a differential scanning calorimetry (DSC) plot, the one at right may well interest you. And even if you don’t, you can see those lively spikes in heat flow and appreciate what that’s going to mean if you have a decent-sized reaction going. (The inversion around 20C is, naturally, the melting point of DMSO – it’s always been a good indicator of a cold lab!) Whether or not you experience the second spike after the first will be a function of how large a reaction you have going, how well it’s mixed, and how much heat you can transfer out of it (and how quickly). Local heating in such cases can send things into a runaway decomposition.

For an illustration of that, see the photo at right. The left-hand panel is a Hastelloy accelerating-rate calorimetry (ARC) cell, before and after doing a DMSO/NaH run, and the right-hand panel shows the ARC apparatus itself after said assay. You don’t want this. You especially don’t want this when your reaction vessel is not contained within a large calorimeter, but is rather a Pyrex round-bottom in the middle of your fume hood. This was a loading of 4.5 grams of a mixture of 84% DMSO and 9.7% NaH (the rest was the mineral oil from the sodium hydride). In case you’re wondering, that model ARC cell has an average burst limit of about 14,500 psi, which should inspire some serious thought.

But you’re not off the hook with DMF or DMA, either. The temperature where trouble hits is a little bit higher in those cases, but they’re trouble just the same. An analysis of the gaseous byproducts from the DMF/NaH decomposition showed hydrogen, carbon monoxide, methane, and even ethylene/acetylene, which argues for a radical chain mechanism, at least in part. In all cases, the more sodium hydride in the reaction, the lower the temperature where decomposition sets in.

So strongly consider alternative solvents or bases for such reactions, especially if you’re working on any kind of scale. And most definitely don’t heat them up! Most industrial process chemists are aware of these hazards (or should be!), but a lot of academic and industrial synthetic chemists aren’t, or don’t realize how bad the problem is. The fact that these combinations are used pretty often shows that you can get away with them under “ordinary” conditions, but the problem (as always) is that you never quite know when things are going to stop being ordinary – and then it’s far too late to do anything but hope you’re not nearby.

37 comments on “Sodium Hydride in Aprotic Solvents: Look Out”

  1. The Lucky One says:

    The very first chemistry experimental in the very first chapter of my thesis says the following: “To a solution containing 100 mL of DMSO and 200 mL of benzene was added 17.0 g of 60% NaH in DMSO. The reaction mixture was heated at 75 °C for 1.5h…” That would have been recorded for posterity in J Org Chem circa the late 1990’s.

    1. The Lucky One says:

      Yeah, so that would obviously be 60% NaH in mineral oil…

    2. A Nonny Mouse says:

      The best experience of dimsyl sodium was early in my PhD when I went in one Sunday to find a post doc making the stuff (from Somalia who was supposed to have a grant for some agricultural work but was doing chemistry (very occasionally). Also had managed to tell 2 Nobel prize winners to F off- he didn’t mange the 3rd -the head of department- as he thought the same about the other 2!)

      His recipe was to throw solid sodium metal onto undried DMSO- ON THE BENCH- and see the results, which happened to be a 10 foot jet of flame.

      I decided to take the day off.

      1. anon says:

        Hey, with those skills maybe he could have gotten a nice bomb-making job back home.

        1. A Nonny Mouse says:

          I understand that he was jailed for misappropriation of funds (I am sure that most of it was spend on the “ladies” of the Fulham Road from what I hear- he had a room in the student dorm that I was in a few years before)!

  2. Old Timer says:

    That’s why I never liked the distinction between “aprotic” and “protic” since several “aprotic” solvents are “protic” under the right reaction conditions. Just by saying “aprotic”, people shut down their mechanistic mind when thinking about possible side reactions (or explosions, in this case).

  3. milkshake says:

    By the way, another underappreciated risk is reaction of alkali metals or LAH with ethers at high temperatures. An exotherm from reaction of Na metal with boiling diglyme was what led to T2 explosion in Jacksonville

  4. A Nonny Mouse says:

    I usually use NaH/THF and then slowly add NMP up to a max of 5% (you can stop the addition when hydrogen evolution starts to be seen).

    I have had no problems, but I don’t know if they exist or not.

    1. Derek Lowe says:

      The paper mentions a scale-up example using THF as co-solvent, actually. Sounds reasonable. . .

  5. Caltech Chemist says:

    Is NMP a reasonable alternative solvent that does not have this issue?

    1. A Nonny Mouse says:

      Unfortunately, it does, but it is not as bad as DMF/DMA which is why I use it with THF.

      1. milkshake says:

        tetramethylurea and dimethylimidazolone might be

  6. Lazy guy says:

    Why would somebody heat a concentrated NaH solution without substrates in a close system? Oh, never mind…

    1. Nameless says:

      You can argue that DMSO was the intended substrate in this experiment.

    2. ReOrgChemist says:

      because they are NOT lazy and actually care about safety?

      1. Lazy guy says:

        The problem is not NaH/DMF, but the high concentration and heating. You don’t usually need excess NaH and high temperature to does its job.

  7. Nick K says:

    The danger of NaH/DMF seem not to be as widely known in industrial labs as it should be. Many years ago there was a serious explosion in a Process Research lab at Sanofi Montpellier due to a runaway exotherm with this combination. The reaction had been run many times on a kilo scale without difficulties, but one day it got just a little too hot and took off.

    My own experience with NH isn’t happy: on two occasions the material (dry salt and mineral oil suspension) has inflamed spontaneously while I was weighing it out.

    1. Nick K says:

      Typo in my last post: “NH” should be “NaH”, of course.

      1. Thoryke says:

        I was just figuring you were no longer welcome in New Hampshire…. 😉

    2. A Nonny Mouse says:

      I had this once (oil) but it wasn’t due to the NaH, but rather some very active lump of sodium which was in there; made a nice hole in the trespar bench.

  8. Anonymous says:

    I’ve successfully used dimsyl anion from NaH/DMSO (research lab, not process lab) without any dismal results. (Rim shot.) This OPRD paper has reawakened my awareness.

    Very often, chemists think that some compounds or combinations are safe or inert, but that can be relative. I get kind of nerdy and correct junior colleagues who sometimes refer to doing “reactions under an inert atmosphere” and mean dinitrogen (N2), but even N2 is not inert to some reactants or catalysts. I prefer argon anyway, but the nerdiest among you will direct me to the wikipedia entry on “Argon Compounds” . (Argon warning! Ar condenses at 87 K; if you are using a liquid N2 bath, 77 K, your flask or trap will slowly fill with liquid Ar! Watch out for rapid expansion when you remove the LN2 bath! I learned that as a 2nd year: I watched as the liquid level rose in my flask even though I wasn’t doing anything yet. I realized what was happening, vented the setup, let it warm up safely, and started again with N2.)

    THF is used with strong bases but THF is not inert. THF can be deprotonated to release ethylene (gas) and acetaldehyde enolate. (You can’t cleanly prepare acetaldehyde enolate from CH3CHO and strong base (e.g., LDA); elimination with sBuLi or tBuLi is one of the preparative methods.)

    Junior colleagues: talk to your more experienced senior colleagues before trying some chemistry that is new to you. Senior colleagues: keep alert to potential hazards and do not dismiss junior colleagues who ask you questions or for your help when evaluating the safety and proper way to use some new chemistry.

  9. Silicon says:

    Speaking of underappreciated safety concerns. Ever see the paper about trifluoromethylphenyl grignard decomposition ( This is something I never thought of back when I was making Grignards of trifluoromethylphenyl derivatives. In some cases the degradation gave curves with over 2000°C/min and nearly 10,000psi/min pressure changes! Now that’s a hell of a runaway if you ask me. Certainly another good example of a non-intuitive safety concern that could have horrific consequences if ignored.

    1. Nick K says:

      The dangers of aryltrifluoromethyl Grignards are well described in an Org Syn prep:

      Extremely violent detonation may occur if the solvent boils off, causing the Grignard reagent to solidify.

  10. Anonymous says:

    I should have added: Everyone should be aware of and have access to Bretherick’s Handbook of Reactive Chemical Hazards, 8th Edition or later. It is THE most useful safety reference, IMO. (It might be available on-line through your institution. PLACE A DIRECT LINK ON YOUR DESKTOP!)

    Too many chemical safety guides and handbooks are kind of useless to bench chemists. “If you ingest large quantities of KOH, do not induce vomiting. Call a doctor ASAP.” “If you fall into a giant vat of toluene, climb out, remove your clothing, get fresh air and call for assistance.” Worthless to a bench chemist. Bretherick is an indexed listing of actual hazards and warnings, with references.
    I am sure that this OPRD will be incorporated into the next edition.

    Two more important hazards to be aware of in the “real” chem lab world: (1) Did you ever see a smoky vapor from acetone and CHCl3 and wonder what it is? Acetone and CHCl3 react exothermically with even a trace of catalytic base or metal (rusty waste can). Today, most labs segregate halogen waste from non-halogen waste but I will bet that, in the past, hundreds or thousands of exploding waste cans were caused by the acetone – CHCl3 exotherm. (2) Nitric acid plus alcohols, e.g., ethanol. Bretherick calls that mixture “rocket fuel.” I have seen the aftermath of HNO3 – EtOH explosions that happened repeatedly in the same lab.

    You can also look up “Can of Beans” to see what happened with that.

    1. CMCguy says:

      I 2nd the recommendation on Bretherick’s being an invaluable resource in providing warning on potential hazardous reaction and conditions as in fact derailed my plan to scale up a NaH/DMSO process I had worked out in the lab before learned the issues. Found other chemistry there that took for granted in Grad school projects as not being risky that realized were performed close to the edge of disaster with no one had second thoughts about.

      Per above I agree people need to openly discuss and review their experimental plans with others, especially as old timers often have awareness and experience (which unfortunately gets lost when they are let go or lucky to retire). Definitely was common in Process labs to consider risks and always had a Safety assessment before pilot plant operations (and work with skilled Engineers and EHS for added input).

    2. A Nonny Mouse says:

      We have always used ethanol + a couple of drops (COUPLE!!!) of nitric to clean sinter funnels in a good fume hood.

      1. Druid says:

        This is a radical reaction in which nitric oxide and nitrogen dioxide. If you want to avoid the inconvenient induction period, you can start the reaction with a tiny crystal of sodium nitrite.
        An induction period of a radical reaction can be dangerous because the mixture does nothing until you turn your back on it.

    3. Nick K says:

      I knew that mixing acetone and chloroform leads to a relatively mild exotherm (hydrogen bonding), but I was unaware of the danger you describe. Do you have a reference for that?

      Many thanks.

      1. Anonymous says:

        Nick K: If you can access Bretherick, there are primary literature citations given there. Descriptively, CHCl3 anion (-)CCl3 adds 1,2 to the acetone ketone to give Cl3C–C(OH)(CH3)2, chloretone.

        The reaction is exothermic and self-propagating. The non-preparative version can happen in a waste drum. I have no proof, but I think that many decades of waste can geyser eruptions (or explosions) were improperly dismissed (even by Profs and senior PIs) as “Who knows?” or “Probably some free radical reaction.” and so on. I think that CHCl3 – acetone was a likely cause of many of them. I always had the safety seminar speakers (not always chemists!) mention the CHCl3 – acetone hazard and a few others. When the $-cost of waste dispo went way up, departments started to segregate halogen waste from non-halogen waste. That reduced the likelihood of additional spontaneous chloretone syntheses.

        And here’s a variation:

        1. Nick K says:

          Many thanks for this info.

  11. MeanGreen says:

    Dimsyl anion can be a nice deuterating reagent. Add the smallest speck of KOtBu to your NMR sample (DMSO-d6) and you might be surprised where you can selectively deuterate heterocycles, haloarenes, and other interesting sites.

  12. Snark says:

    So… Sounds like you need a “Things I -might- work with” category.

  13. Baltic says:

    The risks of “typically considered aprotic” solvents (a more appropriate name, IMO – with some effort you can make a whole lot of innocuous-looking substances to decide it’s time to give up some protons) in combination with sodium hydride are quite well known, especially to the industrial chemists. Still, a common commercial packaging option for NaH is “in DMF-soluble bags”, meant for “just chuck the whole thing in” applications.

  14. Scott says:

    Only a 15kpsi bursting strength?!? You need tougher lab equipment!

    Say, a decommissioned ‘baseball’ hand grenade, since those have a nice threaded top on them where the fuze goes in. Though you will need to weld up the hole drilled in the bottom to prove there is no explosive filler inside…

    I think you could get that ARC cell to burst under conditions of a steam-line rupture, water going from liquid to gas phase expands some 1700x, right?

    1. e says:

      Uh, I think you need to rethink that. You are thinking about expansion at atmospheric pressure. What temperature would you need to heat water to in a sealed system to generate 15000 psi? I don’t know the answer because the Anntoine parameters I can find don’t even touch those temperature ranges…

  15. Chris says:

    We had a gram scale alkoxide prep in THF as part of a macromolecular substitution process, and the scale-up was contracted out to someone with no proc dev experience. When the process came back they’d swapped the solvent to DMF so they could significantly increase the concentration. We got handed a process with ~100 g of NaH in a limited vol of DMF at high temperature and we didn’t like the look of it. We ran it at smaller scale to check and saw immediate and significant gas evolution before we added the substrate, with both dimethylamine and CO being identified. We switched to diglyme and told our contractor they were lucky to still have a lab.

  16. Lith-iator says:

    Madelung indole synthesis: RLi reagents were not in a routine use at the time in our labs . So let’s think of a strong base, stronger than KOt-Bu, We thought. Let’s test dimsyl sodium then a popular strong base . RT no reaction, somewhere at 80°C , boom but mild. Is dimsyl potassium better or worse

Comments are closed.