Wednesday, June 27, 2012

Process Wednesday: molten sodium? Yeah, we got this.

What does it take to run a reduction of an oxime with 5 kilograms of melted sodium? I'm glad you asked -- Breitenmoser et al. would be happy to answer for you. [1] The reaction scheme is below:

How about making sodium sand so it would be flowable? (I see to recall a Dave Collum JOC procedure to make lithium sand -- I'm glad I never got asked to make that one):
A first familiarization experiment to generate granulated sodium by cooling melted sodium in xylenes below its melting point while stirring ended up with a broken glass impeller stirrer at 750 rpm. A second experiment without stirring during cooling led to sodium plates, which tended to aggregate upon restart of stirring. Therefore, all further experiments were carried out above 100 °C (sodium mp 98 °C) and at approximately 250−500 rpm, avoiding such problems by handling only liquid sodium. No stirring was applied during the melting of sodium. 
Never mind! After lots of RC1 and ARC studies, they finally arrived at a procedure that they were comfortable with. Reactor design became an issue:
To ensure a safe scale-up of the procedure described above, an intermediate scale-up with 410 g sodium (5 L reaction volume) was performed in a 100-L vessel (scale-up factor 90) to test the different reactor geometry (cylindrical Büchi steel-enamel reactor with impeller stirrer and baffle located at 11.5 L, minimal stirring volume 1 L). One change with respect to the run in entry 7 (Table 4) was carried out: only 1 equiv 4M2P (4-methyl-2-pentanol) instead of 2 equiv 4M2P was added prior to the addition of the oxime 2 to increase the excess of sodium by 1 equiv. Possibly, this amount could be further reduced, but this was not studied anymore. A lower than expected conversion of 64% was measured after 5 h at 115 °C. The low volumes led to rather inefficient stirring (baffle with different plates not immersed), meaning that the sodium was not finely dispersed by all baffles. Upon further scale-up, a faster conversion was expected... 
Indeed, in the next intermediate scale-up run with 2.0 kg sodium (23 L reaction volume), a conversion of 91% after only 2 h at 120 °C was measured.
I like this article, if only as a demonstration of the importance of reactor configuration in process development. Since it's a heterogenous reaction solution (melted sodium, starting material, toluene/xylenes), the quality of mixing became one of the main factors in determining reaction rate. Baffle height and volume became key to getting the an acceptable rate.

Finally, it's wonderful to see a situation in which scale-up leads to a faster reaction, as opposed to a slower one. Would that they were all like that.

[1] Breitenmoser, R.A.; Fink, T.; Abele, S. "Safety Assessment for the Scale-up of an Oxime Reduction with Melted Sodium in Standard Pilot-Plant Equipment." Org. Process. Res. Dev. ASAP


  1. on small scale the preferred stirrer for crating molten sodium dispersion for making sodium dust (in molten paraffin, for example) is a tight-coiled flexible thin steel wire spring that is roped into an "infinity sign" loop centered at the end of the shaft. Such flexible stirrer is very abuse resistant (by being bendable) and can be brought to high RPMs required for fine Na dispersions, for example when preparing sodium hydride

  2. I think they call them Hirschberg stirrers.

  3. I've used sodium potassium alloy (NaK) in the past. VERY reactive, but liquid at room temperature and ambient pressure.

  4. It seems that the acceleration in the final scale-up was due to poor vessel fit of the previous process (5L reaction volume in 100L nominal vessel, rxn vol below baffles).
    They may have had to use the vessel that was available.

    The reaction mix includes some sodium alkoxide. Has anyone seen any corrosion of glass lining from Na alkoxide? I am working on a similar problem and there is precious little published data.


looks like Blogger doesn't work with anonymous comments from Chrome browsers at the moment - works in Microsoft Edge, or from Chrome with a Blogger account - sorry! CJ 3/21/20