Solvent Exchange at Scale in the Plant

The need for solvent exchanges in the sense of displacing one solvent by another without passing through a liquid-free state exists only in the plant or pilot-plant settings.  Doing chemistry in the laboratory, when one solvent needs to be replaced with another, the solution contents are placed in a round-bottomed flask, set spinning on the vacuum rotary-evaporator with appropriate heating and strong condensing, and when the first solvent has been completely evaporated then the required new solvent is added and the solutes brought back into solution by swirling and scrapping.

However, at scale, evaporation to dryness is not possible without caking and possibly charring. Even if it were possible to avoid degradation, the layer of non-volatile residue would become so thick on the reactor walls that heat transfer to complete the evaporation would be made impracticable.  Combined with this difficulty, at low volumes in a normal reactor stirring stops.  Thus, solvent replacements must be done without completely removing all the liquid phases at any point.

As an exercise, let us consider solvent replacements among a dozen of the most common solvents.  This examination is an analytical exercise. None of the more complex, multistage switches have been experimentally verified.  The only inputs to these proposals are miscibilities that are known, solvent boiling points, and the data from binary azeotrope tables.

I have named my list Common Reaction Solvents because they are not all solvents of choice for process chemistry. Chloroform, for example, would not be used today in a chemical process, and hexane because of its flashpoint would be questionable. Besides all being affordable together they span a wide range of physical properties. Seven of these solvents are miscible with water. Four are immiscible with water.  Two (ethyl acetate & dimethylformamide) can be hydrolyzed by aqueous acid or aqueous base. Three are dipolar aprotic. None make an aqueous phase either acidic or basic.

Common Reaction Solvents

Methylene chloride               39.6

Acetone                                  56.1                

                                          Chloroform                            61.2                                                

Methanol                                64.5                                                        

Tetrahydrofuran                                66.0                                

Hexane                                  68.7    

Ethyl acetate                          77.1

Ethanol                                   78.3

Isopropanol                            82.2

Toluene                                  110.6

Dimethylformamide               153.0

Dimethylsulfoxide                 189.0

                        

The Number of Combinations

There are 12 x 11= 132 possible binary replacements among this group of frequently used solvents. 

In considering a solvent switch from any first solvent on this list to any second solvent on this list, certain replacements are not a challenge. So long as no azeotrope is formed and so long as the difference in boiling points is more than about 30 centigrade degrees any replacement from a lower boiling solvent to a higher one can be done trivially by simple distillation through a column of a few theoretical plates.

Thus, methylene chloride can be directly replaced by ethyl acetate, ethanol, isopropanol, toluene, DMF, or DMSO. Acetone, chloroform, methanol, THF, hexane, and ethyl acetate can each be replaced by toluene, DMF, or DMSO. Isopropanol can be replaced by DMF or DMSO and toluene can be replaced by DMF or DMSO. Finally, DMF could be replaced by DMSO. 

This takes care of 29 of the 132 switches. We have 103 remaining to consider.  The more difficult replacements will be those from a higher-boiling to a lower-boiling solvent or between the two solvents close together in boiling points. 

When a replacement is done by simple single-stage or azeotropic distillation one does not need to take into account the solubility of the non-volatile constituents of the solution. Whether these solutes are less soluble or more soluble in the second solvent compared with the first does not matter.  Since the solutes we are considering would be relatively non-volatile they cannot escape. They will stay with the liquid phase in the reactor no matter what it becomes.

Some exchanges cannot be performed by distillation alone but require some additional action.

DMSO Replacement

Let us consider first the replacements when DMSO, the highest boiling solvent on our list, must be replaced.  If DMSO is mixed with 'mineral spirits' (pet. ether bp 179-210 C) and the mixture is distilled (preferably under high vacuum), DMSO, as the more volatile of the two liquids, will be the distillate leaving the non-volatile solutes together with the non-polar mineral spirits, dissolved or as an immiscible phase. 

On the other hand, if DMSO is diluted with a little water and mixed with hexane, two phases will form, but non-volatile solutes will only transfer substantially into the hexane layer if they are distinctly lipophilic. Polar solutes will remain in the DMSO-water layer no matter that a choice of phases is available.  Continuous extraction is not really a viable option at scale.  It is too complicated and too slow.

In solvent replacement schemes that comprise some extraction, the polarity of the solute being transferred is critical. In this article, I will use the term ‘somewhat polar solute'. My operational definition will be a 'somewhat polar solute' is one that can, in no more than three equal-volume extractions, be removed in at least 90% yield from a hexane solution into a methanol solution. Such solutes, I imagine, are generally less than 400 MW and contain several at least moderately polar functional groups such as alcohol, ketone, ester, or amide.

Transfer of a 'somewhat polar solute' from DMSO to Methanol

Preparing a methanol solution of such a solute

To transfer a 'somewhat polar solute' from DMSO to methanol one can imagine the following sequence of steps:

1.  Under vacuum, concentrate the DMSO phase to the 'minimum stirrable volume.

2.  Add mineral spirits and co-distill under a vacuum to remove the last portion of DMSO as in the first paragraph of this section.

3.  Cool the reactor contents.

4.  Add methanol to the slurry/oil/solution between the solutes and the mineral spirits.

5.  Stir the two phases to partition the solutes.

Since mineral spirits are mostly paraffin molecules, only the most apolar solutes will prefer that pure hydrocarbon phase.  The 'somewhat polar solutes' will prefer the methanolic phase. Several extractions with methanol should transfer the solute.  Wash the combined methanol extracts with hexane to remove traces of the high-boiling mineral spirits. Alternately, if instead of mineral spirits, a liquid that is completely straight-chain paraffin is used for the displacement, the residual paraffin can be removed from the methanol by adding and then crystallizing urea from the methanol. This will generate a urea inclusion complex, which will effectively remove the straight-chain hydrocarbon from the methanol.  

Note that this idea for cleaning up the final traces of straight-chain hydrocarbons in mixture with other solutes has never been put to an experimental test. The formation of urea complexes with kinds of paraffin alone is treated in Differential thermal investigation of complex formation in the system urea-n-paraffin. A.V. Topchiev, L.M. Rozenberg, N.A. Nechitailo, E.M. Terent’eva, Zhurnal Neorganiicheskoi Khimii (1956) 1, 1185-93. and also the same authors Doklady Akademii Nauk SSSR (1954,) 98, 223-6.  There is also an older article that touches on all the types of substances that can be treated this way, O. Redlich, C.M. Gable, A.K. Dunlop, R.W. Millar. Addition compounds of urea and organic substances. J. Am. Chem. Soc. (1950), 72, 4153-60. 

Of course, this option cannot be used, if the desired solute is a straight chain of more than 6 atoms with no significant branching.  The solute would also get trapped in the urea complexing.

Whatever method is used to remove the last bit of high boiling hydrocarbons, this solution in methanol will be called Solution M because it will be used as a starting point and a first step for other exchanges.

Transfer of a ‘somewhat polar solute' from DMSO to one of toluene, hexane, ethyl acetate, acetone or chloroform

Using Solution M as obtained above, somewhat polar solutes can be transferred to the subtitle solvents because every one of these five form an azeotrope with methanol.  What is important in doing this is to concentrate the methanol solution to the smallest practical volume and then add sufficient of the second solvent to move to the correct side of the azeotropic composition so that a vapour composition richer in methanol than in the second solvent distills causing the composition of the remaining liquid to become increasingly richer in the second solvent at the expense of methanol. 

 The acetone solution that can be obtained when a methanol/acetone mixture is concentrated will be called  'Solution A' because it is used further below.

Transferring ‘somewhat polar solutes’ from DMSO to Ethanol.

The solution in DMSO is converted to a slurry or solution of the solutes in mineral spirits by distillation under a vacuum as described earlier in this blog. Mineral spirits are like hexane and so will form a distinct two-phase mixture with either pure acetonitrile or acetonitrile into which a trace of water has been added. Any ‘somewhat polar solutes’ will shift to this predominantly acetonitrile layer.  Acetonitrile forms an azeotrope with ethanol and so by diluting with an appropriate volume of ethanol and distilling the azeotrope this sequence enables a transfer from DMSO to ethanol.

Transfer of ‘somewhat  polar solutes’ from DMSO to DMF or Toluene 

 The transfer from DMSO to Solution M  achieved above can lead to subsequent transfer into toluene or DMF by simply distilling away the methanol since there both of these have differences in bp of at least 30 C. 

Transfer of ‘somewhat polar solutes’ from DMSO to methylene chloride, THF, or IPA

Methylene chloride

Taking  Solution M as a starting point, methanol can easily be distilled out under reduced pressure replacing it with water.  If the ‘somewhat polar solutes’ are insoluble they will separate as an oil in water emulsion. When all the methanol has passed over, adding methylene chloride will take up the solutes. Cutting the phases and drying the methylene chloride by boiling out the azeotrope of methylene chloride and water will leave the solutes in methylene chloride alone. 

Transfer of ‘somewhat polar solutes’ from DMSO to THF

A transfer from DMSO to THF would seem to be possible starting with a transfer to Solution M which could then permit a transfer to Solution A.  Pentane could be added to a mixture of Solution A and THF. THF is not reported to form any azeotropes but pentane and acetone are reported to form an azeotrope of composition 21.0% acetone and 79.0% pentane which boils at just 32 C and this is well below the bp of THF.  This azeotrope boiling point therefore would be more than thirty degrees lower than the bp of THF so THF and this zeotrope should be easily separated. Once the azeotrope has been taken overhead, any excess pentane can be distilled away from the THF since THF and pentane themselves also still differ by about 30 C degrees.  Note that this scheme also although derived logically from available data, has not been tested in the lab as far as I know. 

The switch from DMSO to THF is the most complicated transfer we have so far considered. An alternative when the solutes are soluble in hexanes might be to displace DMSO with just enough high boiling glyme solvent (diglyme, triglyme, or tetraglyme) to maintain stirring in the reactor after the DMSO has been completely driven off. Then add hexanes to the glyme solution or slurry and add solid calcium bromide. Glyme forms an insoluble complex with excess calcium bromide in hydrocarbon (sometimes even in other non-complexing solvents). The glyme complex can be filtered and the hexane removed and replaced by the THF.

Transfer of ‘somewhat polar solutes’ from DMSO to IPA

The same methodology as was proposed above for transferring 'somewhat polar' solutes from DMSO to THF can also be used to make the transition from DMSO to isopropanol (IPA). 

This completes the examination of solvent switches from DMSO when the solutes are 'somewhat polar'.

Replacing DMSO when the Solute is Apolar

 Suppose the solute is non-polar.  If we blindly use our previous strategy which first codistills with mineral spirits, the apolar solutes will remain in the mineral spirits and not be extracted into either a methanol or an acetonitrile layer, so we would be 'snookered'.

In this situation, we must rather displace the DMSO with a  high boiling polar liquid.  I would be worried to use a protic solvent like glycerine or diethylene glycol because the hydrogen bonding with the DMSO may create a maximum boiling azeotrope, which has not been documented or at least could raise the boiling point by entrainment of the DMSO.  Triglyme or tetraglyme dimethyl ethers are liquids, which might be effective and not entail these risks. Like DMSO they are not proton donors and they are sufficiently high boiling to allow selective volatilization of DMSO.  Since they are miscible with water we can use the strategy of further decreasing the solubility in the glyme dimethyl ethers by diluting with water.  Then we can expect to extract the hydrophobic solute into a low boiling medium like pentane.  From pentane we can replace with any of the solvents from methylene chloride up to DMF (11).

If the solvent were are trying to use as our second solvent is higher boiling than methylene chloride we will have more flexibility and will choose something like hexane rather than pentane, which has too low a flash point for comfort in the plant.

An alternative to using a glyme methyl ether would be to use low molecular weight solid polyethyleneglycol. This material would chase the DMSO easily.  It is really not volatile itself and serves just as a heat transfer medium while remaining liquid and stirrable in the reactor.  When the DMSO has been driven off, addition of hexane or methyl t-butyl ether should precipitate the polymer and allow the solution of solute in the low boiler to be filtered away from the solid polymer.

This would depend upon the polymer not being soluble at all in the second solvents.  Diethyl ether is the usual solvent used to precipitate polyethylene glycol, but it is not welcome in the plant setting.

Solvent Replacements from DMF

The same strategies that we have applied with DMSO would be applicable to DMF and be easier to implement because DMF has a boiling point about 30 C degrees lower than DMSO but still very different from our other selected common solvents. Again, the intermediate solvent, which is used to chase the DMF can be either strongly hydrophobic or strongly hydrophilic to suit the solute.

With DMF another option that becomes more practical as the bp of the first solvent falls is steam distillation. Of course, this choice is not made if the solute is appreciably water-soluble.

An option sometimes is to hydrolyze the DMF with aqueous acid after the DMF solution has been concentrated. If the solute of interest is reactive to strong acid it can be protected by dissolution in another immiscible organic layer during the aqueous acid treatment. The products. dimethylamine and formic acid, are both very water-soluble and are unlikely to solubilize most solutes.

In many cases where large volumes are not a concern, either DMSO or DMF can be diluted with a large amount of water and the mixed aqueous phase extracted with an immiscible organic phase.  This of course cannot be done with very water-soluble solutes.  It is however the most common work-up procedure and runs into trouble only on scale where a high point of maximum volume limits the throughput.  This is important in the early steps of longish processes.

Replacements from Toluene to Solvents of Lower Boiling Point

Isopropanol

Isopropanol and toluene form an azeotrope bp 80.3 C containing 5.0% IPA and 42.0% toluene. The two solvents also form a ternary azeotrope with water bp 76.3 C which can be used to dry the IPA after the solvent replacement.

Ethanol

Ethanol also forms a useful azeotrope with toluene with bp 76.7C and composition 68% ethanol d 32% toluene. Again there is a ternary azeotrope with water bp 74.4C.

Ethyl Acetate

Ethanol can be replaced by ethyl acetate utilizing an ethanol/ethyl acetate azeotrope bp. 71.8; 31.0% ethanol and 69.0% ethyl acetate.

Methanol

Methanol also forms a useful azeotrope with toluene with bp 63.7 C; composition 72.4% methanol and 27.6% toluene.

Ethanol forms an azeotrope with hexane of bp 58.7C and composition 21.0% ethanol and 79.0% hexane.

Chloroform forms an azeotrope with methanol bp 53.5 C; composition 87% chloroform and 13% methanol.

To replace from toluene to methylene chloride use the water azeotrope with toluene to remove all the toluene and give a water slurry, then extract back into methylene chloride after saturating the water with salt to increase the extraction efficiency.

 Water immiscible solvent to a Second Water immiscible solvent

Transferring from any water-immiscible solvent with bp less than 120 C going to another water-immiscible solvent can be done by adding acetic anhydride as the chase solvent, removing the first solvent, then adding water and acid or base to hydrolyze the acetic anhydride followed by extracting the product into the new water-immiscible solvent. Functional groups that are acetylated by acetic anhydride can be used so long as the hydrolysis conditions return the unchanged solute. December 16/2011.

The same method can be used to take the solute into the water-miscible solvents THF and acetone. Once in water saturate the water with salt and extract. The saline will give two phases with either acetone or THF but this method is not clean and gives very wet solvents.

We have not achieved a good replacement between toluene and either acetone or THF.  If we could find a replacement to acetone we could use the acetone/pentane method to get into THF.

Toluene can be taken into methanol and methanol into acetone; then acetone pentane can be removed from solution with THF to replacement to THF.

Replacements from Isopropanol to Lower Boiling Solvents

An azeotrope exists between IPA and EtOAc with bp 74.8 C and composition 77% ethyl acetate and 23% IPA.

An azeotrope exists between isopropanol and hexane with bp 61 C abd composition 78% hexane and 22% IPA. Another azeotrope exists between Ethanol and Hexane with bp 58.7C and composition 21% ethanol nod 79% hexane. Using these two azeotropes in sequence one can do a replacement from IPA to Ethanol.

The azeotrope between IPA and Hexane can be used to make this replacement.

The replacement from IPA to THF

IPA to Hexane; Hexane to Methanol; methanol to acetone; mix acetone with pentane and THF and distill acetone/pentane azeotrope leaving THF.

IPA to hexane; hexane to chloroform

IPA to hexane; hexane to methanol; methanol to acetone

IPA to water; water extract with methylene chloride

Replacement from Ethanol to Ethyl Acetate

Ethanol to hexane; hexane to methanol; methanol ethyl acetate

Replacement Ethanol to Hexane by an azeotrope

Ethanol to THF

Ethanol to water using toluene in a ternary azeotrope;

Ethanol to Methyl cyclopentane

Once a solvent change has been made to give a solvent with a bp of 60-70C there is no further incentive to change solvents.  Solvents with a bp, of 50-70 do not leave difficult-to-remove residuals when crystalline solids are separated from them either at atmospheric or reduced pressures. The reason for solvent replacements among solvents in this boiling range is simply to find a solvent from which the solute recrystallizes in a pure form, a high recovery, and desirable crystal morphology.

An acetone solution can by azeotropes replacement into bromopropane, acetone carbon tetrachloride, 1-chloropropane, cyclohexane, diethylamine, iodoethane, isopropyl ether, methyl acetate,

From chloroform one can do a replacement to methyl ethyl ketone,

From ethanol one can move to the solvents acetonitrile, benzene, carbon disulfide, carbon tetrachloride, 1-chlorobutane, chloroform, cyclohexane, dibromomethane,1,2-dichloroethane, diethyl formal (diethoxymethane), ethyl acetate, hexane, heptane, isopropyl acetate, isopropyl ether, methyl ethyl ketone, nitromethane, tetrachloroethylene, toluene, trichloroethylene, triethylamine,

Ethyl acetate ca be exchanged for the following pure solvents: carbon disulfide, carbon tetrachloride, ethyl acetate,

Hexane can be used to exchange for pure solvents: 1-butanol, t-butanol, isopropyl ether, methyl ethyl ether, t-butanol, nitromethane, n-propanol,

Isopropanol can lead to other pure solvents: cyclohexane, diisopropylamine, ethylene dichloride, methyl ethyl ketone, tetrachloroethylene, trichloroethylene,

Methanol can lead to pure solution through azeotropes: acetonitrile, benzene, cyclohexane, 1,1-dichloroethane, dimethoxymethane, dimethylformal, ethylenedichloride, ethyl formate, heptane, hexane, isopropyl acetate, methyl acetate, methylal, nitromethane, octane, trichloroethylene, trimethyl borate,

Toluene can be used to provide pure solvents: acetic acid, t-amyl alcohol, 1-butanol, 1-chloro-2-propanol, epichlorohydrin, 2-ethoxyethanol, ethylenediamine, glycol, isobutyl alcohol, 2-methoxyethanol, nitroethane, nitromethane,1-propanol, pyridine.