Using Chasers in the Work-up to make a High-boiling Solvent Practical

Importance of Solvent Choice

In the reaction of two generalized chemical species, A with B, the most significant variable invariably affecting the yield is the stoichiometry. The optimal stoichiometry is usually something close to the molecular proportions in the balanced chemical equation representing the desired reaction.  Certainly, these are the proportions that chemists hope will be best because any excess of either substrate will cost extra money.

Frequently, the second most significant variable is the choice of solvent. This makes intuitive sense. In an uncatalyzed reaction, the solvent is the only other chemical next to the reactants that is present within the transition state and it is the reduction of the transition state energy that makes the conversion to desired product preferred over unreacted starting materials or byproducts.

The more solvents to choose from, the greater the opportunity to improve a reaction’s selectivity. Of course, some solvents are preferred on the basis of cost per liter. Others are preferred for ease of removal in the isolation and purification of the product. Usually, this is because it is too high boiling/viscous for the work-up or makes drying tedious. Whatever it is, because solvent choice is a discrete variable, (if binary mixtures are off limits) there is a widespread tendency to quickly settle for one of those ‘old faithfuls’ and then systematically work with the continuous variables to ‘optimize’.

It is true that if 

• the substrate and reagent (A and B) are both soluble in a solvent and 
• that solvent has already been successfully used in an example in the literature 

your chance of successfully adapting it is substantially increased;

however, there are cases where benefits can accrue by looking beyond the ‘old faithful’ solvents. Those benefits are most likely to be realized

(a) if there is plenty of room to improve the reaction’s yield 

(b) if a good ‘in situ’ assay for the desired product is available, and 

(c) if one knows how to conduct an efficient search.

On the other hand, there are good reasons for not considering these less-utilized solvents.

 Reactions that cannot be totally quenched deteriorate during solvent switching. For example, if the substrate can over-react with excess of a reagent but excess reagent is necessary to give the required conversion, then tacking on a solvent switching operation to get rid of a higher-boiling reaction solvent, will most likely lead to overreaction and extra byproduct formation. Similarly, if the desired reaction product is thermally unstable, the extra heat input and time spent for solvent switching may prove deleterious.

 But if any excess reagent can be first completely destroyed or otherwise disabled, solvent switching to separate a higher boiling reaction solvent can still be considered.

It is not necessary to demonstrate the separation from a higher boiling reaction solvent unless such solvent actually seems to be delivering the required improvement in reaction yield. At this scouting stage in an ‘optimization’, improvement only needs to be hinted at by an improved assay for the desired product in the completed reaction mixture. This is why having a dependable product assay needs to be in hand before looking for a less-common reaction solvent. 

Switching from High-boiling Reaction Solvents for Work-up

A reaction solvent can be removed by distilling it away from an even higher-boiling solvent called the chaser solvent. I will consider five different chaser solvents: Acetic anhydride, Quinoline, Triethanolamine, PEG 400 (liquid polyethylene glycol), Glycerin, and Paraffin.

Each of these chasers has some unique feature that allows it to be in turn easily exchanged for something low-boiling to continue the isolation and purification.

 Acetic anhydride can only chase solvents of boiling point less than 140 C. It works because it can be converted to an aqueous acetic acid-water mixture that will be immiscible with many low-boiling, classic organic solvents that may be preferable for separation, purification, and isolation.

Quinoline is high boiling. It can be removed by steam distillation, even under vacuum for greater stability of the solutes. Trace residues can be removed by extraction with acidic water since quinoline is mildly basic.

Triethanolamine is very high boiling and can be a chaser for even rather high-boiling solvents. It is miscible with water. Traces that are carried over to a new lower boiling solvent can be extracted with aqueous acid.

PEG 400 is essentially nonvolatile. It can be a chaser for any organic solvent. It can be precipitated with diethyl ether.

Glycerin, although very viscous, can occupy the minimum stirrable volume in a reactor and allow another lower boiler to be distilled away. Glycerin is immiscible with a wide variety of regularly used organic solvents. Glycerin will keep polar solutes in solution.

Paraffin is the opposite polarity extreme to glycerin. It is essentially non-volatile straight-chain saturated hydrocarbons. It can occupy the minimum stirrable volume in a reactor allowing a reaction solvent to be completely replaced. Because paraffin is made up of long-chain hydrocarbons, it is immiscible with regular solvents which are themselves immiscible with hexane, heptane, cyclohexane, etc. Traces of paraffin, because they are straight chains in structure, can be removed as urea inclusion complexes that crystallize from methanol.

Mixed Xylenes as a Possible Extraction Solvent that is Immiscible with DMSO, DMF, and Trichloroethylene (TCE)

 In a chemical process step, the unseparated mixture of positional isomers of xylene is cheap enough to serve as either a reaction solvent or solvent for use in purification.

I am always on the lookout for pairs of organic solvents that can serve as immiscible phases for solute partitioning by liquid-liquid extraction since this is a very robust, simple, and scalable purification method.

Although toluene is immiscible with wet DMSO, it is miscible when thoroughly dried. However, the commercial xylene mixture is reported to be immiscible with even dry DMSO. This mixture of positional isomers also is reported to give two liquid phases with dimethylformamide and trichloroethylene. The extra saturated carbon apparently makes the difference. 

Of the three combinations:

xylenes/DMSO

xylenes/DMF

xylenes/ trichloroethylene

the final one seems the most remarkable.  I would appreciate it if someone who is actually in a lab (I am retired) would either confirm or disavow it in the comment section. It would be very interesting to see how different compounds are partitioned between these two.

Second Crops of Crystals are Easily Available from a Gas-Expanded, Mixed-Solvent System

 One of the advantages of performing crystallization of a substrate from a single solvent by cooling as opposed to causing crystallization by diluting a first solvent with a miscible anti-solvent is that one can try for a second crop simply by reducing the volume of the filtrate, recool the reduced volume to yield more solid. One can do this because the solvent composition isn't being modified. This advantage would be retained if the crystallizing solvent is a lower-boiling binary azeotrope.

In the alternative, where an anti-solvent is being mixed in to create the required supersaturation considerable tedious work is required to remove all the anti-solvent and concentrate that first pure solvent before a second crop can be attempted.

But if the anti-solvent is a gas under plant conditions, this re-establishment of a single solvent and its concentration is simple. Take for example a mixed-solvent recrystallization that was originally being performed by dissolving the substrate in toluene and then decreasing the overall solubility by adding hexane and then cooling. Suppose instead one dissolves the substrate in toluene cools the solution but instead now bubbles in butane gas. The butane will dissolve in the toluene but the solubility of the substrate will decline in just the same fashion that occurs by adding hexane. The product will crystallize. You cannot filter using a vacuum since this would drive off the butane. Filtration must instead be done by pushing the slurry through the filter cloth with pressure. When the crystallized substrate has been caught on a filter, evacuating the system will easily remove the butane from the filtrate leaving the toluene which can be further concentrated. A second crop can be isolated by repeating the gas expansion with butane.

Furthermore, although mixed solvents are not normally recycled and reused in multi-purpose fine chemical plants, Gas-expanded liquids are an exception since simple distillation rather than fractional distillation is sufficient to do the job.

Any mixed solvent recrystallization that uses cyclohexane, hexane, heptane or petroleum ether can be rejigged as a gas-expanded liquid mixed solvent recrystallization using butane thereby enabling taking a second crop of crystals to raise the yield.

Formamide: an Organic Reaction Solvent from which Product can be Easily Recovered

 Formamide is a clear, hygroscopic, oily liquid, miscible with water, methanol, acetone, acetic acid, dioxane, ethylene glycol, glycerol, and phenol. 

It is dried with solid sodium sulfate or calcium oxide.  Activated alumina is also reported to be suitable for drying formamide. High-purity formamide is vacuum distilled and packaged under dry nitrogen. 

Many compounds such as tannins, starch, lignin, polyvinyl alcohol, cellulose acetate, and nylon, dissolve in it. It also dissolves many ionic compounds that are insoluble in water, making it a great solvent for salty/sugary reactions.  Chlorides of copper, lead, zinc, tin, cobalt, iron, aluminum, nickel, and the acetates of the alkali metals as well as some inorganic sulfates and nitrates dissolve. Zinc chloride, stannous and stannic chlorides, ferric and ferrous chlorides, aluminum chloride, and copper chloride are all organic chemistry reagents that might consequently benefit if used in formamide.

The Leuckart reaction necessarily uses formamide since it is required also as a reactant. Formamide should be considered a replacement for reactions more commonly conducted in water. It is polar, protic, and both a hydrogen bond donor and acceptor. It has been used in microwave-assisted syntheses. It degrades in a microwave at 170 °C to CO & NH3 which may be done deliberately as a source of reactant CO or NH3.

Similar to water, at temperatures close to 200ºC hot formamide begins to dissolve a wide variety of commonly functionalized organic compounds and thus can be used as a solvent for organic reactions. At 200ºC these reactions run exceptionally fast. When cooled to room temperature, the organic products become practically insoluble and can be easily separated. Unlike many organic solvents, but similar to water, formamide does not deteriorate at high temperatures. As a result, the formamide filtrate could be repeatedly used as the solvent in the same reaction.  

Its high BP and decomposition to small amounts of HCN at reflux lead chemists to often prefer dimethylformamide (DMF). Neither formamide nor DMF is easily removed with a vacuum, but formamide has the lower solubility in benzene/toluene compared to DMF so it's possible to remove an organic reaction product from formamide by benzene extraction. Extraction with toluene or xylene would also be expected to be OK although it is not documented.

BP: 210°C

Density: 1.133 g/mL

Avoiding Product Loss in an Aqueous Layer During an Extractive Work-Up of an Organic Reaction

 

Most organic reactions are quenched by adding an aqueous phase into the reaction vessel followed by some appropriate workup protocol. At some point, the water-rich phase is cut away and usually discarded since usually the molecule of interest is more soluble in some hydrophobic organic solvent. If one hopes to achieve a good ‘isolation yield’ the aqueous phase must not retain any product. This may require back-extractions of the aqueous layer depending upon the solubility which in turn depends upon the number of carbons in the product and the functional groups that comprise it. Since a standard reactor is emptied through a valve in its bottom, multiple extractions of an aqueous layer are most convenient when the organic layer that will hold the product is denser than water and is the layer closest to the bottom valve. Unfortunately, this is rarely the case. Most organic solvents are less dense than water; that is, the aqueous phase is at the bottom of the reactor and if more than a single extraction is to be performed on the aqueous layer, an extra vessel is needed for the operation. 

It would be useful to know whether multiple extractions of an aqueous layer are likely to be required at the time a process step is being planned. This can be estimated roughly by tallying the balance between hydrophilic and hydrophobic fragments in the anticipated desired product.

 We can get a rough idea of what the balance is between hydrocarbon elements and different functional groups by looking at the solubility of some small organic molecules in water. For example, N-butanol is not miscible with water while 2-methyl-2-propanol (t-butanol) is. All the simple pentanols give two separate layers when mixed with water suggesting that one hydroxy balances four to five carbons worth of a substrate.  For the ketone functionality, acetone is completely miscible with water while methylethylketone will lead to two phases so the carbonyl is sort of balanced by three more carbons. The hydrophilicity of polyethyleneglycol suggests the number two for the balancing carbons for an ether link and three may be an approximate number to balance an ester group. An amino group would need about 6 carbons to balance it. A free carboxyl would also be balanced by about 6 carbons.

What am I getting at? If you were to try to prepare for example 7-(carboxypropyl)-18-hydroxy-13-oxa-10-oxo-octadecanoic acid and needed to extract it completely from an aqueous phase, it would be wise to extract with at least two portions of your organic extractant since this molecule is likely to have its hydrophilicity and hydrophobicity close to being balanced giving appreciable water solubility. 

Let us look at a different example for which we actually have some quantitative experimental data. Herbert Feltkamp and Wolfgang Kraus published [Ann. 651, 11-17 (1962)] a study of the liquid-liquid partitioning of 6 possible stereoisomers of decahydro-1,4-naphthalenediol. Now, if as I have suggested, one hydroxyl function’s contribution to hydrophilicity would be balanced by about five saturated carbons then these decahydronaphthalendiols should have significant solubility in both water as well as an organic extraction solvent. In fact, this is what the authors found. 

Although it is not pertinent to the point being made here, because none of these stereoisomers distributed themselves overwhelmingly either in a water phase or a 2:1 v/v mixture of ethyl acetate and petroleum ether (the organic phase they used) the authors were able to separate the isomers by a series of liquid-liquid partitioning extractions!

Why might it be useful to know in advance that several organic extracts might be needed to clear a substrate from the aqueous phase? Suppose we are trying to isolate some product from a reaction mixture. We have taken this molecule of interest into an aqueous phase either as a complex or a salt and we then have broken the complex or neutralized the salt and intend to extract it back into a lighter-than-water organic phase. From the structure of our desired product, we might be able to correctly predict that is still going to have significant solubility in water and may require several organic extractions to essentially recover it quantitatively. We now want to extract that molecule back into an organic layer. But if our molecule of interest retains some water solubility we wouldn’t be able to quantitatively take it up willy-nilly with just one extraction using any old organic solvent. 

We can do one or both of three things:

We can select an organic extractant that will dissolve the rather polar substrate very well. Heptane or cyclohexane would not be good choices but amyl or isobutylene alcohols might work well 

or 

and this is the choice that is likely to be most often overlooked, we could add inorganic salt into the aqueous layer to salt out that product so the fraction partitioned into the organic layer is very substantially increased so that only one extraction is necessary to take up essentially everything. This is the more flexible expedient because, if it turns out that just one extract is needed, the salting-out can be scrapped without any other change in your isolation/purification plan

or

we could take up the organic solute by multiple extractions using an organic solvent denser than water such as methylene chloride, chloroform, trichloroethylene, chlorobenzene etc.

Whatever the situation with which you, the process chemist, are confronted you should have anticipated how extractions are likely to proceed and have planned so that the number of transfers and vessels dirtied are minimized so as to minimize costs.

Acetic Acid-Cyclohexane a Solvent System that can be either one or two phases!

Acetic acid and cyclohexane are two very different substances that are nevertheless miscible above 3.9℃ (their UCST). In a cooled reactor, these will form two separate liquid layers allowing for liquid-liquid partitioning of any solutes therein. More easily than in the laboratory, in the plant the reactor can be kept closed and inerted; consequently, more easily water-free. The acetic acid in the lower layer will be glacial acetic acid so long as water is neither introduced into nor produced in the procedure. More polar components of a reaction conducted therein might be removed by simple phase separation. Glacial acetic acid will be much better at dissolving some substances than a mixture with water present to any extent.

After a cut, there will still be some acetic acid residue in the cyclohexane layer, but acetic acid and cyclohexane give an azeotrope bp. 79.6 ℃ that contains 2% acetic acid. Thus, the predominantly cyclohexane layer can be freed of even traces of acetic acid by distilling out from the reaction vessel a small first fraction. A work-up is possible, still without adding any water!

Where Could Such a Solvent System Be Useful?

This cyclohexane/glacial acetic acid solvent mixture, above 3.9℃ when it is a single-phase, might be a good candidate for conducting acetylations with either acetic anhydride or acetyl chloride reagents. The excess reagent might be removed without decomposing it by cooling to <0℃ and cutting the two phases.

This mixture of fluids might also serve for dehydrations or acid-catalyzed rearrangements. Adding anhydrous hydrogen halides would protonate acetic acid, giving rise to a very strong acid in situ. Excess hydrogen halide would be removed with the acetic acid-rich layer when the reactor was cooled. The system would protonate olefins perhaps inducing rearrangement but hydrogen halide would be unlikely to add across the unsaturation since the halide anion would be strongly solvated and deactivated by hydrogen bonds with the acetic acid.

Acetic acid might catalyze enol formation from ketones. Enols could react internally with a terminal double bone to give a cyclic product or they could be condensed, dehydrated, and so dimerized.

This post is speculative. It does not report experimental evidence.  

The Utility of Ether Solvents and Special Auxiliaries with Organometallic Reagents

Organometallic reagents, such as organolithium compounds and Grignard reagents, are not monomeric in solvents that cannot donate Lewis electron pairs; therefore, because they are self-associated they often are not as reactive with an organic substrate as the monomer would be.  Ether solvents, with or without other chemical auxiliaries, can often dissociate organometallic reagents into monomers without degrading them. Each useful ether has its own limited temperature range. At too low a temperature the solvent either solidifies or becomes too viscous to be worked with and at too high a temperature reaction between the organometallic and the ether destroys the reagent. These de-oligomerizing solvents vary in such important aspects as price per unit volume, ease of purification, and the simplicity with which they can be made effectively anhydrous. The extent to which each of the solvents is miscible with water and recoverable from water also plays a role in the practical usefulness of each.

Special Auxiliary Chemicals

Besides ethers and polyethers there are various other complexing additives that can be useful when added in some low multiple of the organometallic’s molarity. These also modify the reactivity and stability of an organometallic reagent already in one of these ethers.  Sometimes, one of these additives makes it possible to use an ether solvent that otherwise would fail when used alone. Sometimes,  one of these additives activates a reagent so effectively that a solvent that is even less complexing than an ether or not complexing at all (like toluene) can be employed successfully. Some of the additives used successfully are: N,N’-tetramethylethylenediamine; HMPA; N,N’-dimethyl- ethylenephosphoramide; triethylenediamine; or lithium bromide.

Because ethers have such useful properties in organometallic reactions and can be supplemented or replaced, associated special advantages and disadvantages are important to know. Solvent choice is in general the most important discrete, discontinuous element among reaction conditions ( time and temperature are continuous variables). Let us look at the various ether solvents.

Diethyl Ether

Diethyl ether is the most frequent solvent for making Grignard reagents in the laboratory. Butyllithium can also be prepared in diethyl ether in the laboratory. At scale diethyl ether’s dangerous flammability and its exceptional tendency to form peroxides makes it unsuitable for making organometallics of any kind without extraordinary costly precautions. It is simply not done. 

Tetrahydrofuran (THF)

THF is the most common replacement at scale when making both Grignard and organolithium reagents. When the combination of too high a temperature with too long a time organolithiums decompose THF.  For example, a molecule of butyllithium splits THF into one equivalent of ethylene and one equivalent of the anion of acetaldehyde. 

THF has the disadvantages that it is miscible with water and consequently is a problem to recover and it forms no azeotropes to help in solvent switches to remove water and to dry it.

2-Methyltetrahydrofuran

2-Methyl THF has become popular because it has the advantage of being largely immiscible with water, thus enabling its simpler recovery while retaining the ability to complex organometallic reagents. Its boiling point of 80.2ºC is still acceptable but of course, it is more expensive than THF.

Arylboronic Acid Functionality makes Phase Switching Possible

 

In the thesis of Paul O’Brien working with Professor Organ at York University in Toronto, there is useful information for using molecules containing aryl boronic acids as intermediates that enable a phase switching purification.

 This thesis teaches that phenyl boronates can be extracted into a water solution containing 1.0M D- Sorbitol at pH 11-13 and then after making the pH acidic, the boronate functionalized substrate can be taken back into a water-immiscible organic solvent, thus achieving a solvent switch that can provide a rugged tool for rapid purification.

“Hall demonstrated that aryl boronic acids are stable to reaction conditions such as IBX oxidations (2-iodoxybenzoic acid), DIBAL-H reductions, esterification of alcohols, amidations, Wittig reactions, and Grignard additions [Mothana, S.; Grassot, J.; Hall, D. G. Angew. Chem. Int. Ed. 2010, 49, 2883-2887].This strongly suggests that arylboronic acids may be stable to a range of rather harsh reaction conditions without resorting to protecting groups. 

Arylboronic acid tagging strategies also show good atom economy in contrast with other tagging methods because the tag can be utilized in chemical transformations such as Suzuki cross-coupling, Rhodium cross-coupling (C-N and C-C), chemoselective oxidations, catalytic hydrogenations, selective aerobic oxidative coupling, transition-metal free cross-coupling, and protodeboronation to name just a few.

The use of arylboronic acid moieties as phase-switching handles adds synthetic utility while reducing synthetic steps such as protection protocols for subsequent column chromatography. Purification of most boronic acids is difficult since they are highly polar, like carboxylic acids, and adhere to polar chromatography material such as silica. The degree of lipophilicity of the boronic acid substituent influences chromatographic adhesion, where highly lipophilic substituents decrease retention on silica. Another method to decrease the polarity is boronic acid protection using 1,2-diol such as catechol or pinacol esters to mask the polar -B(OH)₂ functionality as a lipophilic ester.”

I have made what I think are editorial changes in what is almost, but not quite, a quote.