Solvents that can be Dried Using the Water Azeotropes

 

In a previous blog article, the point was made that drying organic solvents with solid desiccants would contaminate the entire filter train and thereby drastically increase cycle time. Drying using an azeotrope was much preferred. All of the common organic reaction solvents that are not completely miscible with water form azeotropes that can be used for drying. It is only essential that the solute is stable at the azeotrope’s boiling temperature.

Only the common solvents completely miscible with water cannot be dried by azeotropes:  acetone, methanol, tetrahydrofuran, dimethylformamide, dimethyl sulfoxide, and propylene carbonate. 

Solvent Water 

Azeotrope

Acetic acid  76.6

Acetonitrile  76.5

Benzene  69.4

1-butanol  93.0

2-butanol  88.5

t-butanol  79.9

Chlorobenzene   90.2

Chloroform  56.3

Cyclohexane  69.8

1,2-dichloroethane  72.0

1,2-dichlorobenzene 91.1

Ethanol  78.2

Ethyl acetate  70.4

Ethyl ether 34.2

n-Heptane  79.2

n-Hexane  61.6

Isopropyl acetate  75.9

Isopropyl alcohol  80.4

Isopropyl ether  62.2

Methylene chloride  38.8

Methyl ethyl ketone  73.4

MTBE  52.6

n-Pentane  34.6

Propanol  88.1

Tetrachloroethylene  88.5

Toluene  85.0

Trichloroethylene  73.1

m-Xylene  94.5

Binary Azeotropes of Common Organic Solvents Potentially Useful for Crystallization

 

 Example 1of US patent 3,932,384 awarded to Sawa et al. reads:

 

“The resulting crude base (17.4 g) is passed through a column of 87 g of silica gel and eluted with benzene to yield the pure base (15.58 g) which on recrystallisation from an azeotropic mix of benzene/n-hexane affords 15.64 g (93.2%) of 6-benzyl-3,4-dimethoxy-10,11-methylenedioxy-5,6,7,8,13,14-hexahydrodibenz[c,g]azecine….”^#

^# There must be some error in their yield calculation

In 2009, when I searched, this was the only recorded example of recrystallisation from an azeotrope, with the exceptions of those using the commercially available azeotropes between water and ethanol, hydrogen chloride, or hydrogen bromide.

In a previous blog article, {https://kilomentor.blogspot.com/search?q=azeotropes#google_vignette}, KiloMentor proposed six azeotropes that might usefully serve as recrystallising solvents. Here, I want to look at a broader combination of common solvent possibilities.

In a binary lower-boiling azeotrope, each component of the solvent mixture reduces the cohesion of the molecules of the other component. As a result, the vapour pressure of the azeotropic mixture is greater than the sum of the partial vapour pressures of the components added together. Each component is repelling the other into the vapour above the mixed liquid itself. Consequently, the combined vapour pressures match atmospheric pressure at a lower temperature than either component boils separately. 

If a binary mixture of solvents with the composition of a lower-boiling azeotrope were used to dissolve a practical amount of an essentially nonvolatile solute, such as would be required in a recrystallisation, and the mixture refluxed, the azeotropic mix of solvents might not distil over. The azeotrope might ‘break’.

 

The nonvolatile solute is likely, to some extent, to disrupt the original vapour-liquid equilibrium (VLE). A soluble solute almost always alters the relative volatilities of the two original components differently due to varying intermolecular interactions, which cause the composition of the vapour to no longer be identical to that of the liquid at the original azeotropic point. There is an uncertainty in using an azeotrope versus a single pure solvent for a crystallisation. 

Indeed, some nonvolatile solute will cause the relative volatility between the two original components to become so large that the azeotrope disappears from the phase diagram. This turns the dissolving fluid into a simple mixture of the two solvents so that boiling will eject the more volatile of the two preferentially.

The addition of a solute sometimes can break an azeotrope if that solute interacts with and depresses the vapour pressure of one solvent component more than the other. A mixture of a solute dissolved in an excess of a lower boiling azeotropic composition cannot be concentrated without risking a change in solvent composition.

Indeed, the combination of a practical amount of a solute with a practical amount of a lower-boiling binary azeotropic mixture of solvents may cause phase separation if the solute there is enough solute and if it has a sufficiently overwhelming preference for interacting with one of the solvent components. Although phase separation may not occur when hot, the solute may oil out in a phase with one of the azeotrope’s components upon cooling.

Despite these unfavourable possibilities, one may be able to predict situations when a substrate has a structure that is likely to interact advantageously with a low-boiling azeotropic mixture of solvents during recrystallisation. A compound that has one portion of its structure that is preferentially solvated by one component of the azeotrope and another domain preferentially solvated by the second component is unlikely to break the azeotrope or cause phase separation. Furthermore, by the same reasoning, such a compound is likely to be more soluble in that solvent mixture than in either of its components alone.

Because a greater weight per volume of an amphiphilic substrate should be soluble in a lower-boiling azeotrope than in the same volume of either of its pure components, a better recovery of crystalline solute might be expected upon strongly cooling it.

Many potential solutes have apolar and polar/hydrophilic and hydrophilic domains within their complete structure, and it is these that are more likely to be usefully recrystallised from lower-boiling binary azeotropic mixtures of common organic solvents such as those below.

 Azeotropic Solvent Mixture          b.p.

5% ethanol in dichloromethane 39.0

7% methanol in dichloromethane 37.8

3% methanol in MTBE 52.6

31% water in MTBE 51.3

33% cyclohexane in acetone          53.0 

41% hexane in acetone          49.8

12% methanol in acetone 55.7

7% ethanol in chloroform 59.4

13% methanol in chloroform 53.5

31% THF in methanol 60.7

20% isopropanol in methanol          64.0

46% hexane in THF 63.0

3% 1-butanol in hexane          67.0

4% allyl alcohol in hexane 65.7

22% 2-propanol in hexane 61.0

21% ethanol in hexane 58.7

23% isopropyl alcohol in ethyl acetate 74.8

31% ethanol in ethyl acetate 71.8

49% methanol in ethyl acetate 62.1

28% toluene in ethanol          76.7

30% isopropanol in 2-butanone 77.3

34% ethanol in 2-butanone 74.8

33% 2-propanol in cyclohexane 68.6

2% acetic acid in cyclohexane 79.7

31% ethanol in cyclohexane 64.9

33% 2-propanol in cyclohexane 68.6

42% toluene in isopropanol 80.6

12% water in isopropanol 80.4

28% acetic acid in toluene 105.4

Another way of looking at this is that these binary mixtures may be particularly good choices for mixed solvent crystallisations, no matter what the proportions.

Finally, some binary azeotropes show particularly large deviations from ideal behaviour. That is, their mixtures show the largest boiling point depression. These, quite possibly, can dissolve significantly more of an amphiphilic solute and give a higher recovery on cooling than other choices. 

Selected Azeotropes with Big Bp Depressions

!st Solvent (bp) 2nd Solvent (bp) Az. bp. Dep.

Cyclohexane(81) Isopropanol (82) [67] 14

Cyclohexane (81) Ethanol  (78)  [65] 13

1-Chlorobutane( 78) Ethanol (78)  [66] 12

2-Chlorobutane(69) Methanol (65)  [53] 12

Ethanol (78) Hexane (69)  [59]         10

Hexane (69) Isopropanol (82) [61] 8

Chloroform (61) Methanol (65)  [54]         7

Hexane(69) Nitromethane (101) [62] 7

Acetonitrile (82) Isopropylether (67)  [62]   5

t-Butanol (83) Hexane (69) [64]  5

1,2-Dichloroethane(83) Methanol (65) [60] 5

Hexane (69) 2-Butanone (80)  [64] 5

 Selection Criteria

• Greater than 10% of the minor component
• Azeotrope bp. <70 C
• 5 Cº degrees or more difference between the lower boiling component and the azeotrope

Preferred Solvent for Synthesizing a Compound that can be Reversibly Extracted into Water.

 

Suppose I were presented with the problem of synthesising an organic compound that can be reversibly extracted into water at some pH (ie, an acid or base). Further, suppose that there is no powerful reason to use a particular solvent for the transformation, but I do know the plant reactor in which the reaction will likely be run in. What would my dream solvent be?

I know the minimum storable volume of the proposed reactor and the maximum volume of the reactor. So I would use a two-component solvent made up of o-dichlorobenzene and hexane. The ratio of hexane to o-dichlorobenzene would be (1/2 maximum volume - minimum stirrable volume) ÷ Minimum stirrable volume. 

Why o-dichlorobenzene? I want to be able to use enough o-dischlorobenzene to fully occupy the minimum storable volume in case it is necessary at some point to distil away all the hexane. Also, o-dichlorobenzene has a density of 1.31, so it will constitute the lower phase if I need to do a liquid-liquid extraction after distilling away hexane. Also, o-dichlorobenzene has a boiling point of 180 C, so it can serve as a chaser if I need to distil away any other solvents with significantly lower boiling points.

Why hexane? Hexane will ensure that the reaction temperature cannot, without resistance, go above 68.7 °C, the boiling point of hexane. I do not know what the optimal reaction temperature will be. A hydrocarbon cosolvent gives me the greatest number of very similar alternatives with different boiling points, which can serve as alternate stable reaction temperatures. Why the particular hexane: o-dichlorobenzene ratio? When I scale up, I want the o-dichlorobenzene to fill the minimum storable volume, while the reactor itself will be only half-filled in case I need to add more volume or perhaps an aqueous layer.

Hexane would be no more than my first choice for the second component of the solvent mixture, however, because any, preferably but not mandatory, water-immiscible solvent that can be distilled away from o-dichlorobenzene might function. I use a mixture of cosolvents because that makes the solvent composition adjustable during optimisation, knowing that waste solvents in the pharmaceutical and fine chemicals businesses are rarely recovered. 

The essential elements of this methodology for making a water-extractable product are (i) to use one cosolvent with a density greater than water and a boiling point distinctly different from that of the second solvent, and in a volume sufficient to occupy the minimum stirrable volume of the planned full-scale reactor and (ii) a final solvent that is immiscible with water.

The plan would be to conduct the chemical conversion at the best time-temperature combination studied. Then, optionally distil away the hexane, and add a water phase at the preferred pH to extract the desired product. Then drain away the lower chlorobenzene layer that contains any organically neutral & soluble co-products, byproducts, reagents or starting materials through the reactor’s bottom valve. Subsequently, any lighter than water organic solvent immiscible with water can be added into the reactor, and after adjusting the pH appropriately, the product will switch back into this upper organic layer, and the water can be drained off to waste. The product will remain partially worked-up, potentially in a solvent from which it can be crystallised or precipitated!

Building Organic Chemistry Synthetic Processes Retrosynthetically

To develop a synthetic process the chemist needs a sequence of chemical reactions and their work-ups, isolations, and intermediate purifications that can be scaled up.  

This is obvious. 

What is less obvious or unrecognized is that the sequence should most advantageously be planned so that the final intermediate before the transformation that gives the target final product, should be one which can be flexibly purified by some rugged methodology, so  that

1. Off-spec material can be upgraded and

2. New impurities introduced by changing prior steps in the reaction sequence will not make it into the final product, where they could change the product’s characteristics, require new impurity identifications, change the analytical methods that have been developed, or require, most critically, that clinical trials be repeated.

Although the functionality in this last intermediate can be anything that can meet this readily purified criterion, having an ionizable acidic and/or basic functionality is the most predictable way to achieve this. However, many targets will not have such a functionality, though most pharmaceutical products will, since salt forms are most desired for making oral tablets.

In US 6204383B1 issued to Torcan Chemical Ltd. for a new synthesis of Sildenafil, the route was planned so that the basic nitrogen functionality was inserted at the end of the synthesis, and the final intermediate was a neutral species.

In fact, the patent reads in its introduction:

 

“As a relatively complicated synthetic organic chemical molecule, sildenafil requires a multi-step chemical synthesis. Any organic synthesis step, which is part of a complex multi-step synthesis, results in contamination of the intermediate with solvents, catalysts, starting materials, and by-products, and so introduces the requirement for purification. If a pharmaceutical grade of final product is to result, this cleaning must be done either as the contamination is caused, that is, in the work-up of the particular step, or at some subsequent point in the process. A rugged process is desirable, which is not demanding with regard to the purity of the intermediates and which allows for a very efficient cleaning during the isolation of the final drug product.

It is an object of the present invention to provide novel processes for preparing sildenafil that simplify the purification procedures and produces sildenafil in substantially pure form without involving complex purification procedures.

It is a further object of the present invention to provide intermediates useful for the preparation of sildenafil by such novel processes.”  

The reason this patent’s stated objective so matches my own thinking is that I was the company patent officer who wrote those words and conceived the strategy!

When there is no appropriate functionality to simplify purification in the product, it would be very advantageous if the last step removes one that has been built into the last intermediate! This makes reactions that remove an acid or basic group from a molecule very important.

The chemist, working out his scheme retro-synthetically, can replace the ultimate target with a penultimate target, which is the target molecule with such a removable acidic or basic function appended!

An example of this can be found in my US43574730A, a patent for making omeprazole. Omeprazole is neither acidic nor basic; that is, it can’t be extracted reversibly into an aqueous liquid layer by adjusting the pH of the water.  I set as my actual synthetic target, therefore, an intermediate that had a carboxylic acid bound to the omeprazole substructure in a fashion so that the final transformation would be an easy decarboxylation. This intermediate had the advantage that, before decarboxylating to give omeprazole, this precursor could be phase shifted into water, leaving any non-acidic materials, whether starting materials, byproducts, or unreacted starting materials, behind in the organic phase. Then, adjusting the pH, this intermediate would transfer back into fresh organic solvent and be decarboxylated to give clean omeprazole. The idea worked just as contemplated. In practice, it turned out that a preliminary secondary amide was the best way to work with the extra carboxyl functionality.

Preparing a Resolution Standard for Use to Validate the Analytical Method for an Active Pharmaceutical Ingredient (API)

 

No analytical method can be proven able to identify all potential impurities in a sample. What can be demonstrated is that a refined method can separate more impurities more completely from the pure substance than others. To develop such a method one needs an impure sample of the desired product containing as many of the process’s potential impurities as possible. This material is called the resolution standard. Besides being used to choose an appropriate analytic protocol, this material is used to show that the analytical equipment is functioning as required by the protocol before samples are run.

How to Make a Resolution Standard

Batches of synthetic API that are quite evidently impure, made by the envisioned process, using whatever unvalidated methods have been used in process development, are selected. As large a sample of each batch is taken as can be sacrificed, and these samples are combined, mixed, and as large a portion as will fit is packed into a mechanical blender. An anti-solvent is added to cover and completely moisten the solid, and the mixture is blended thoroughly but without substantially raising the temperature of the mass. The blending is stopped, and the mass is left for several hours. The wet mass is filtered to remove all the triturated solid. The clear filtrate was concentrated to dryness under vacuum, keeping any warming to a minimum. This solid from the evaporation will be a material with intensified impurities. If a further increase in the proportion of impurities is desired, the same procedure should be repeated, triturating this sample identically.

The method is called swishing.    

A Proposed Cheap Replacement for Quinoline -3-carboxylic acid as a Reversible Tag

 

In 1999, Hélène Perrier and Marc Labelle [J. Org. Chem. 1999, 64 2110-2113 ] proposed using a 3-quinolinecarboxylate protecting group on a growing intermediate substrate to assist in isolating them at each step in a process by precipitating these substances as insoluble quinolinium salts.

The proposal seemed promising but at scale the cost of that protecting group inhibited adoption.

The KiloMentor blog has always aimed to promote and enhance methods that streamline isolation, purification, and workup procedures in synthetic organic chemistry.

In this particular instance, KiloMentor wishes to propose an inexpensive synthesis of a related 6-methyl-3-pyridinecarboxylic acid that might serve in the same way but that could be much cheaper to prepare.

The chemical literature already teaches the high-yield, simple, scalable synthesis of 2-methyl-5-aldehydo-4-pyridone in 75% isolated yield. [ F. Arya, J. Bouquant J. Chuche, Synthesis Communications 1983, 946-948 ] by continuous hot tube pyrolysis from the inexpensive inputs isopropyl amine, ethylformate and dimethylmalonate. 

The conversion of this 2-methyl-5-aldehydo-4-pyridone to  6-methyl-pyridine-3-carboxylic acid has not, as far as I know, been reported; however, it would seem that it might be made simply by the addition of the compound gradually to strong aqueous alkali to oxidize the aldehyde, reduce the ketone (Cannizzaro like) and then dehydrate the ring to aromatize the moiety. The resultant amino acid should be easily separated because of its acid-base properties.

KiloMentor Process Development Strategy Reviewed

 

KiloMentor | revised 6th  January 2009 republished February 17/2017 and March 22nd 2025

This is a revision of one of the earliest articles from the KiloMentor archives. The original was written in 2007.  It restates for new readers the core idea of the KiloMentor process development philosophy and teaches an approach that KiloMentor thinks leads consistently to valuable ideas. for improving process throughput.

In synthesis, we talk about assembling, building, or constructing a molecular structure. This is a misleading metaphor because we are comparing an activity in the nano-world to an activity in the macro-world. Operating in the macroscopic world, for example in building a house, we handle the pieces, we position the pieces, and we join the pieces.

In chemical synthesis, we do none of these. The substructures we are endeavoring to unite are atomic in scale: too small to touch, to align, or even to see. 

In chemical synthesis, the chemist adjusts macroscopic conditions: solvent ratios, stoichiometry, stirring, temperature, duration of exposure, etc. Then the chemist presents the proposed reaction partners, to each other under the orchestrated conditions and they interact, as their nature dictates; but, hopefully, this is also as we have planned.  How is this perspective different from the conventional one?  Chemical process development is simply making these parameter choices that cause nature’s choice to comply with what we want the outcome to be, efficient. Nature to be commanded must be obeyed.

According to the academic synthetic chemistry tradition, synthetic accomplishments are scored based on the number of synthetic steps, the yield per step, and the overall yield for the combination of steps. High yields are good. A short sequence is good. The combination is elegant. According to this traditional perspective, the focus is on the reactants, the plan for reactant transformation, and the overall yield output from that plan. Separation of unreacted starting materials, by-products, co-products, catalysts, solvents, salts, and other excipients and processing chemicals are in the background (the attitude is that work-up/purification can be done and will be done BUT these are not important criteria to evaluate the quality of the synthesis).  The give-away phrase of those who harbor this philosophy is “the product was isolated in the usual way.”

From the KiloMentor perspective, in this age of online substructure searching, coming up with creative transformations with strong literature analogies is no longer the domain of the synthetic genius but has come within the scope of good synthetic chemists. We do not have to depend upon our neuronal computers alone anymore. Now it is creative ideas for separation and purification that are not easy to search for and have become the greater artist skill of the project. The deconstruction of the chemical soup and the fishing out of the desired product in an adequate state of purity is paramount. 

Is there any particular value in this way of looking at processes that surpasses the traditional way which focuses on the series of chemical reactions while taking the separation of intermediates as an obvious technical work? My perspective emphasizes: 

• The work involved in setting up and controlling the necessary reaction conditions. 
• The work involved quenching the reaction condition/then working up the reaction and finally isolating and purifying the desired product. 

The value in this perspective is that in chemical synthesis, the money, manpower, and resources consumed during the reaction step phase, ie. while A & B are reacting with each other, is minuscule compared to the money, manpower, and resources expended preparing for the reaction and recovering pure product from the reaction. 

The clash of these perspectives leads to the question, “Which would I rather do- a four-step synthesis in which every conversion has many parameters that must be rigorously controlled and from which each intermediate must be isolated by gradient column chromatography and evaporated to a foam OR an eight-step synthesis which is rugged and forgiving of process deviations and from which each intermediate can be cleanly extracted in a separatory funnel or crystallized or distilled to give an adequate practical purity intermediate."

People have personal preferences and this is as it should be in a pluralistic society BUT I pick the second sequence and as the need for larger quantities and higher quality intensifies, I increasingly prefer the second route. 

Please note- I am not saying the number of chemical steps doesn’t matter. I am not saying that the overall yield does not matter. I am saying that elegance also encompasses simplicity, ruggedness, time economy, and scalability. 

OK, so what. How does this insight change our behavior in the synthetic laboratory, office, or library? Based on an examination of what really goes on in a chemical process step a method of rating the difficulties of the separation is proposed as a quantitative tool to rank the challenges of a process scale-up.

We should evaluate or rate synthetic schemes using more criteria:

1. Number of Chemical Steps
2. Isolated overall Yield
3. Yields of the Individual Steps.
4. Difficulty Rating for Each Reaction Mixture Separation

The fourth point comprises the new insight. How could we execute this new difficulty rating? We could classify work-ups:

A. The product can be separated practically pure by simply liquid-liquid extraction (ie acid-base pH or other phase switching)

B. Product can be separated by crystallization of precipitation as a filterable solid.

C. Product can be separated by atmospheric or vacuum distillation based on a projected difference in boiling points (based on molecular weights)

D. Product can be separated based on chemical reactivity (formation of reversible simply separable derivative, or destruction of contaminant by reaction)

E. The product seems likely only to be separable in practical purity by chromatography.

Clearly, as process chemists, we want to face more A-C separations and fewer D-E type separations.

The KiloMentor blog will highlight methods to augment isolations and purifications so chemists can improve their ability to assign these ratings and take them into account when designing synthetic chemical processes that can be readily and ruggedly scaled up into the plant.