Getting Free from Dipolar Aprotic Solvents at Scale

Obtaining a Reaction Product Free from Polar Solvents: Efficient extraction of highly polar solvents from reaction mixtures.
following the late Dr. Phillip Hultin

 Dr. Philip Hultin was a professor at the University of Manitoba, Canada who passed away in 2018. Although I did not know this man, the following article, which I present with a few edits, is valuable for process chemists scaling up extractive work-ups. Searching Philip Hultin extraction in Google will provide the original unedited article.

The Problem of the Standard Practice

The most common workup for reactions conducted in DMF or DMSO is to drown out (dilute) with a larger volume of water, extract with a solvent such as ether or dichloromethane, and then wash the organic phases numerous times with water.  The shortcoming with this is that dipolar aprotic solvents also have significant solubility in organic solvents and they can also behave as phase-transfer agents drawing a significant amount of the desired reaction products into the aqueous layer.  The whole process can become very time-consuming with multiple extractions and back-washings.

A more effective way, in the laboratory setting, to remove these solvents is a series of separatory funnel extractions that somewhat mimic liquid-liquid partition chromatography.  Indeed, it is a counter-current extraction but wherein only one phase is moving.  The method retains all the lipophilic materials,  does not create large volumes of waste, and reduces the time required to achieve essentially complete separation of dipolar aprotic solvents such as DMF or DMSO from reaction mixtures.

The procedure uses one larger separatory funnel and three or four smaller ones. For laboratory-scale extractions it is convenient to have a rack that can hold all the separatory funnels in a row.  For “large scale” extractions where the first separatory funnel is larger than 250 mL capacity, stands with rings are more appropriate. 

The Laboratory Procedure

1.  After quenching the reaction and diluting with enough ether to easily dissolve the expected products everything is poured into the large funnel #1 and the reactor flask washed with a little ether. There, it is diluted gradually with a generous amount of water. As the water is added two phases start to separate. Doing the addition of liquids in this order may reduce or eliminate the amount of agitation needed to equilibrate the layers; furthermore, the effect of further small increments of water on the volume of upper and lower phases can be more easily assessed and this correlates with the partitioning of the dipolar aprotic solvent.
  2.  Into each of the remaining funnels #2 through #5 is placed a smaller portion of ether.
3.  Funnel #1 is shaken and allowed to settle. 
4.  The aqueous (lower) layer is run out into funnel #2. 
5.  Funnel #2 is shaken, and while it settles, an additional portion of water is poured into #1 and shaken.

6.  The aqueous layer from #2 is run off into #3, the aqueous from #1 is run into #2, and more water is added to #1. 
7.  All funnels are shaken and allowed to settle.

8.  The process of running the aqueous layers into the next funnel in sequence is repeated until all funnels have been shaken with water and the first aqueous portion resides at the bottom of funnel #5 having passed through each of the ether layers.
9.  It is then run off into the beaker. 
10. The remaining aqueous layers continue to move through the funnels and eventually into the beaker as well.

When this sequence is finished, all the ether solutions have been washed five times with water, and all the water washes have been back-extracted as well. If the remaining ether layers are analyzed by TLC, it will likely be observed that reaction products are in funnels #1 through #3 or maybe #4.  Ether layers that contain products are pooled and can be processed further or dried and evaporated.  In general, NMR analysis of the crude material will not show any sign of residual DMF or DMSO after this treatment.

What is happening here is this: when the quenched mixture is initially extracted in funnel #1, most of the polar solvent goes into the water layer.  Some products and ether are also partitioned into the water, and some polar solvent remains in the first ether layer.  The aqueous layer moves into funnel #2 where it encounters fresh ether.  This extracts products out of the water, and it may also take up some polar solvent.  The process is repeated in each funnel.

But now consider the subsequent water washes.  When the organic layer in funnel #1 is washed with water, the residual polar solvent is extracted.  The actual amount of this polar material is relatively small since most of it went off with the first portion of water.  Thus, much less if any of the desired products are partitioned into the water.  This wash moves through the subsequent ether layers, removing polar solvent from each.  Each successive water wash moves polar solvent forward through the funnels, but as the amount of polar solvent in the earlier funnels drops the ability of each water layer to remove the desired product is reduced too.  The result is essentially an “elution” of the polar solvent with “retention” of the less-polar organic products in the earlier ether layers.

In general no more than five water washes are needed although in certain situations more may be required.

The process can be adapted to the use of solvents that are heavier than water as well.  If the extraction solvent is dichloromethane, the stationary phase becomes the water.  Funnel #1 is charged with the aqueous quenched mixture and dichloromethane, and smaller portions of water are put into funnels #2 through #5.  Dichloromethane portions are passed in sequence through the funnels and collected in the beaker having had the polar solvent washed out.

Modification for Working at Scale

The laboratory procedure described above would be unworkable at scale because too many vessels are required. The goal should be to retain the benefit of multiple extractions and multiple backwashes while reducing the vessels to just two. This could perhaps be modified by using one organic water-immiscible solvent denser than water and one that is less dense with a final combining of these two and recovery of the more volatile by fractional distillation. Using dichloromethane as the more dense liquid and isopropyl acetate as the less dense one the procedure might look as follows:

1. Dilute the organic reaction mixture with 1 part of isopropyl acetate and 3 parts water. 2. Stir and separate the lower water into a stirred tank.
3. Backwash the water with 1 part methylene chloride and return this to the mix with the isopropyl acetate in the reactor vessel.
4. Discard the water plus dipolar aprotic solvent from the holding tank.
5. Remove the methylene chloride by distillation from the reactor.
6. Add 3 parts of water to the residual isopropyl acetate containing the desired organic products and stir and settle.
7. Draw off the water containing the remaining dipolar aprotic solvent.
The products are retained in isopropyl acetate in the reactor.