Monday, October 27, 2014

A safety letter on TMS-azide

This week's C&EN has a safety letter from the University of Minnesota's Taton group that had an accident with TMS-azide that I've covered here. Don't miss the follow-up by Neal Langerman below: 
We recently conducted a synthesis of azidotrimethylsilane (TMS-N3) that resulted in an explosion, significant damage to the reaction hood, and injuries to a student researcher. Although it is still not entirely clear what caused the explosion, it seems likely that the reaction and isolation conditions generated hydrazoic acid (HN3) that detonated within the reaction flask. We write to recommend extra precautions when conducting larger-scale syntheses of TMS-N3. 
TMS-N3 is commonly synthesized by reaction of chlorotrimethylsilane with sodium azide and isolated by direct distillation of the TMS-N3 product from the reaction solvent and insoluble NaCl by-product. We had previously followed the original procedure described by L. Birkofer and P. Wegner (Org. Synth. 1970, DOI:10.15227/orgsyn.050.0107) using dimethyl ethylene glycol solvent, as well as modified versions using other solvents such as di-n-butyl ether (Synthesis 1988, DOI: 10.1055/s-1988-27481). We were reproducing a previously reported synthesis (Bioorg. Med. Chem. Lett. 2013, DOI: 10.1016/j.bmcl.2013.10.004) using poly(ethylene glycol) (PEG, Mn = 300) as the reaction solvent and conducting the reaction at roughly twice the scale described in these previous reports (to generate ~200 g of product). 
The reaction mixture had incubated overnight and was being gradually heated in a distillation apparatus for the purpose of distilling the trimethylsilyl azide product. We observed that magnetic stirring had stopped and that the suspended salts had settled to the bottom of the reaction flask. When the student researcher reached into the hood in an attempt to adjust the distillation apparatus, the reaction mixture detonated. 
We do not know what caused the explosion, but there are many possible explanations. The explosion hazard of azide-containing compounds has been the subject of previous safety letters in C&EN and other publications, and many of these warn of the explosive hazard of hydrazoic acid that may be generated from proton sources. We used a newly opened bottle of PEG as the solvent, and although the supplier data indicated that the PEG was dry, PEG itself is protic and can lead to the formation of hydrazoic acid. It is also possible that unreacted azide salts that had settled to the bottom of the still were overheated to detonation when the stirrer failed. 
Given our accident, and the potential for hazard in the synthesis of TMS-N3, we encourage researchers to take special precautions in carrying out any large-scale preparation of TMS-N3 by any method. We recommend researchers follow these procedures: Reduce the scale of the synthesis so that any possible detonation can reasonably be contained; use mechanical stirring to ensure better heat transfer throughout the heterogeneous mixture; and test the apparatus, solvent, and reagents for moisture. We are extremely fortunate that the student has recovered from his injuries, but we are also convinced that those injuries could have been avoided if these practices had been followed in our lab. 
T. Andrew Taton and Walter E. Partlo
University of Minnesota, Twin Cities
This incident was discussed within the American Chemical Society Division of Chemical Health & Safety shortly after it happened. Although I am disappointed that Taton and Partlo have not identified the direct cause of the explosion, I concur that the generation of HN3 is a likely culprit. “Bretherick’s Handbook of Reactive Chemical Hazards,” entry 1310, discusses the potential of this chemical to detonate and other possible mechanisms.
The authors should add the CAS Registry Number to the chemical name (4648-54-8). If they are not planning a full published incident report, in the Journal of Chemical Health & Safety, for example, then they should discuss the underlying causes. 
Neal Langerman
San Diego
I think it's great practice to report these sorts of incidents to the larger community. (FWIW, I think the "horses, not zebras" explanation is hydrazoic acid.) 


  1. They should have known better than run the reaction in non-methylated PEG straight from the bottle. (And on big scale to hoot). TMS-azide reacts promptly with exotherm with 1.1eq. of iPrOH, this is how I was making tetramethylguanidinium azide in situ, from guanidine free base and TMSN3.

    I would encourage anyone intent on scaling up the stuff to make sure the used TMSCl is free of HCl (by redistilling it from CaH2) and the used solvents are dry. Bu2O is quite nice that it is cheap and you can get it pretty dry just by fractional distillation at normal pressure.

  2. I think the first mistake in this incident was using Bioorg. Med. Chem. Lett. as a reliable source for preparing TMS-N3. This journal would be the last place I would go to do this kind of work. I also doubt that it is necessary to incubate the reaction overnight. TMSCl and NaN3 should react almost instantaneously with gentle warming. In fact, the reaction should be set up for distillation right from the beginning because TMS-N3 boils at such a low temperature relative to the solvents being employed here. Giving this reaction time to bake is only giving HN3 more time to generate in larger quantities.

  3. Perhaps a more thorough safety review of this process would indicate whether this prep can be done outside of a blast proof enclosure. I would not want to use an oil bath and an RBF for this reaction, either. A jacketed reactor with a circulator and a secondary overtemp shutoff would be my minimum setup.

    Also, using the reaction mixture prior to distillation a an equivalent of TMSN3 solution perhaps could be explored.

    I wonder if BHT (solvent stabilizer) is a strong enough acid to cause issues with HN3.

    1. For 200 g scale they decided to use a magnetic stir bar. It would be pretty tough for even a large stir bar to move all the NaCl being generated at this scale. That's probably why they discovered that the stir bar was not spinning in the morning.

  4. PEGs are known to contain peroxides (DOI: 10.1002/jps.20726). They usually are more or less stable at rt, but begin to decompose at elevated temperature, yielding hydroxyl radicals. Azides can react with radicals (DOI: 10.1021/ja01017a035 and DOI: 10.1002/chem.200802710). So, for this reaction I'd better use something, that can be distilled from sodium metal prior to synthesis, thus destroying and removing those peroxides. PEG-300 obviously can not be purified this way.

  5. I wonder if it would be beneficial to have an acid scavenger present in the reaction mixture to neutralize any HN3 if it is generated. Maybe something like BaO or CaO?

    1. Both Ba and Ca azides are less stable than NaN3. However, this line of thought reminded me that Teflon coated stir bars can crack and expose the heavy metals of the magnet to the reaction mixture. The exposure would result in highly explosive heavy metal azides.

      Could this be a mechanism of the original incident?

  6. I'm missing something here: wouldn't the TMSCl react first with the terminal hydroxyl groups of the PEG?

    1. Typical procedure requires use of a tertiary amine for TMS protection of alcohols. Imidazole is probably most commonly used, IIRC.