Wednesday, March 27, 2013

Process Wednesday: how long to deoxygenate 1000L IPA/water with N2?

Don't know how long it takes? In the middle of a very interesting article about homocoupling problems in Suzuki reactions, Miller et al. [1] busts out some very interesting, yet very basic measurements on dissolved oxygen in isopropyl alcohol / water solutions and removing that oxygen (emphasis mine):
Having established the important role of dissolved oxygen in promoting formation of dimer 6, we sought to directly measure* and identify an upper maximum limit for dissolved oxygen. Three one-liter-scale experiments were completed in order to assess the efficiency and feasibility of deoxygenation using subsurface sparging.  
In these experiments, a mixture of water and 1-propanol was saturated with oxygen by subsurface addition of air, resulting in measurement of 8.3, 8.7, and 8.6 ppm oxygen. Nitrogen gas was then introduced at a rate of 0.5 standard cubic feet per hour (SCFH) resulting in reduction of dissolved oxygen to less than 0.5 ppm in 2.5 min. Thus, at 0.5 SCFH for 2.5 min, a total of 0.02083 standard cubic feet (SCF) of nitrogen was required to deoxygenate 1 L of solution. On this basis, we calculated that for 1000 L of reaction a purge rate of 30 SCFH of nitrogen for 41.7 min would be required to reduce the oxygen level from saturation to less than 0.5 ppm. Measurement of the recovered solution volume indicated that solvent loss due to evaporation was approximately 0.8%. We concluded that deoxygenation via subsurface sparging would be efficient and practical for large-scale work.
*An ICM model 31250 dissolved oxygen probe and meter was used for laboratory work. Devices such as Metler-Toledo InPro 6800/6900 or InPro 6800 Gas may be found more suitable for manufacture of clinical material.
Ultimately, the authors end up doing this in their sub-kilo reaction (5.8 L solvent, 800 grams of SM) and getting a 93% yield. The concerns that I see with doing subsurface sparges of nitrogen on scale (apart from the loss of solvent from evaporation (problem should not scale?)) is the possibility of aerosolizing material and losing it to upper parts of the reactor.

Nevertheless, a really basic question answered nicely. (There's even a procedure for the experiment!)

[1] Miller, W.D.*; Fray, A.H.; Quatroche, J.T.; Sturgill, C.D. "Suppression of a Palladium-Mediated Homocoupling in a Suzuki Cross-Coupling Reaction. Development of an Impurity Control Strategy Supporting Synthesis of LY451395." Org. Process Res. Dev. 2007, 11, 359-364.


  1. I think that if your reactor is cold enough the formation of aerosols should not be a problem. Solvent loss can easily be avoided by running your condensers at -25°C. The last time I did one of these we did not bother removing dissolved oxygen and everything worked just fine. Their process must be very sensitive to oxygen.

  2. I have degassed large scale reactors by simply bubbling N2 via the bottom values and never bothered about possible evaporation losses. Similarly also have used vacuum cycling technique by applying and releasing in short bursts several times appeared to work just as well and was easier to implement in a few tanks that had appropriate valve configurations