The Dramatic Effect of Moisture on Process Steps & Important Methods for Drying Reagents and Solvents

 

It is known that the outcome of many chemical reactions is sensitive to trace amounts of water and this water often can arise as an impurity delivered from reagents and solvents. It is also known that removing the last vestiges of water from many solvents and reagents is both very difficult and time-consuming and furthermore, measuring low levels of residual water can be difficult. It is also true that in theory, it should be easier to dry and keep dry, larger amounts of materials in larger vessels than smaller amounts in smaller vessels. One reason for this is that the ratio of the surface of the containing vessels to the volume of these vessels is smaller as their size increases. Thus, for example, if water is adsorbed onto the wall of the vessel and from there contaminates the liquid, the amount of contaminant per unit volume of liquid will be smaller the larger the vessel.

A process chemist should certainly want to know what the effect of different levels of trace water would be on his process steps because this is a factor that can have profound effects but might still be within the accepted variance specification of his reagents or solvents. Put another way, the process chemist needs to know the effect of trace amounts of water in inputs so (s)he can set proper analytical specifications for the pilot plant or kilo lab.

Suppose the process chemist plans to use a reaction that is already known to be affected in terms of the yield and/or quality of the product by trace levels of moisture. That is, the moisture content of reagents and solvents are critical variables. Then, in costing the reaction step, the expense of buying such grades of inputs or of drying a cheaper grade to the higher specification must be included. In addition, consideration must be provided for the additional in-process testing that will be required to be assured that the correct anhydrous state is being maintained and to be aware of the heightened risk of process failure from this source. A reaction might prove just too persnickety to use if the requirement for being anhydrous is so strict.

High sensitivity to trace amounts of moisture increases risk. One of the ‘usual suspects’ that needs to be ‘rounded up’ when a process step is found to be irreproducible or that is out of control for no apparent reason, is the level of trace water in the reactor. One way to reduce somewhat such potential variation is by rigorous and effective drying of reagents and solvents. If a reaction behaves the same in terms of yield and product quality when performed with super-dry or regular reagent-grade solvents, then solvent moisture is most likely not going to be a critical variable. The drying of solvents rather than reagents is by far the more common difficulty because solvents are used in so much larger amounts compared with the reacting chemicals.

For drying any particular solvent the best drying agent and method of use of that agent has not been successfully predicted and needs to be empirically determined. The analysis to assess the residual trace level of water in the solvent is not simple so some literature guidance is valuable. David R. Burfield has written a series of papers that are authoritative on this question.

Acetone

Acetone is one of the trickiest solvents to dry thoroughly. The reason is that many of the agents that frequently are used to dry solvents cause condensation of acetone which produces water as a by-product. The preferred drying agent for acetone is boric anhydride. Using stirring and sequential drying, this agent gave less than 18 ppm of moisture in the acetone.

Acetonitrile

Acetonitrile is also most preferably dried with boric anhydride although this is not a particularly good agent for other solvents. The acetonitrile should be first distilled to remove gross water.

p-Dioxane

Calcium hydride is shown to be a good and rapid siccative for dioxane and this is assisted more by stirring the slurry rather than refluxing it. Calcium chloride is also a rather good desiccant, which is somewhat surprising in light of the fact that p-dioxane forms a strong complex with anhydrous calcium chloride.

Toluene

Burfield has not performed experiments with toluene; only with benzene. Toluene is expected to be, like benzene, an easy solvent to dry since water is immiscible with it. With benzene, azeotropic distillation followed by removing 20% of the benzene charge significantly dried benzene but nowhere near as dry as using an effective desiccant. This is important to recognize since distillation of a portion of solvent would be a preferred approach for drying at scale but would not be comparable in effectiveness with a chemical treatment.

Dimethylsulfoxide

Dimethylsulfoxide can be dried by combining it with an immiscible hydrocarbon and codistilling the mixture. When heptanes are used, for example, an azeotrope of heptanes and water distills and the heptanes can be returned to the still pot, and the water separated. In this way, the level of water can be reduced to about 0.2%. To proceed further one can fractionally distill through a packed column with variable takeoff and reduce the water content by another factor of 10 by discarding about 20% of the charge.  Further drying with 4A molecular sieves can reduce the level to 10 ppm.

https://patents.google.com/patent/US6939962B2/en

“Many solvents are capable of removing water by co-distillation, and any solvent that can achieve this can be used in the method of the present invention. Preferred solvents are those which are immiscible with water and form a constant-composition minimum-boiling azeotrope with water. Exemplary classes of solvents are saturated hydrocarbons, aromatic hydrocarbons, chlorinated hydrocarbons, ketones, and ethers. As for the polar aprotic solvent, safety and toxicity considerations will also affect the choice of an appropriate solvent. Hydrocarbons are preferred, and cyclohexane, n-heptane, toluene, and isooctane (2,2,4-trimethylpentane) are particularly preferred The most preferred solvent is cyclohexane.

The codistillation solvent is added to the mixture substantially continuously and is conveniently also added at a substantially constant rate. However, changes in the rate of addition and even stopping the addition altogether for a portion of the reaction are also possible.

The prior art EP-0776903-A prescribes that the solvent capable of removing water by codistillation be added in gaseous form. That is, the solvent, typically cyclohexane, is preheated, and vapors of the solvent are added to the reaction mixture. While this procedure can be used in the method of the present invention, it has been found that it is also possible to add the solvent directly as a liquid, without pre-heating. Typically, the solvent will be added through a polytetrafluoroethylene (PTFE) tube, into the mixture, preferably close to the agitator of the reaction vessel. It will be appreciated that, under these conditions, the solvent will then vaporize rapidly.

It is preferred to recover and re-use the codistillation solvent. This can easily be achieved by condensing the vapors from the reaction vessel and then washing and drying the condensate. More preferably, the codistillation solvent is recycled during the reaction. This can be achieved by condensing the codistillation solvent into a reservoir during the reaction and using that reservoir as the supply for the codistillation solvent to be added to the reaction mixture. In the case that the polar aprotic solvent is DMF and the codistillation solvent is cyclohexane, it has been found that it is best to add a small amount of water to the reservoir. This is because a small amount of DMF co-distills from the reaction vessel with the water and the cyclohexane. The DMF can dissolve in the cyclohexane in the reservoir, and thereby solubilize water in the cyclohexane. When water is already present in the reservoir, most of the DMF is taken into the aqueous layer, reducing the solubility of water in the cyclohexane, and thereby affording drier cyclohexane.”

Dimethylformamide

Type 3A molecular sieves used sequentially are the best drying agent. Phosphorus pentoxide is the best agent that operates by the reaction. Phosphorus pentoxide probably phosphorylates the oxygen of DMF and water is destroyed as it is hydrolyzed. Codistillation with heptanes or cyclohexane as for DMSO would also work but the level of drying is not reported.

Methanol

Water is the most significant impurity in methanol. Exposure of methanol to air increases the moisture content of each exposure. Methanol is often used as the drying agent for the walls of large-scale reactors. Clearly, the extent of drying there will be limited by the water content of the methanol.

Burfield and Smithers have noted that in contrast to ethanol for which there are many warnings in the literature that improperly dried ethanol will give lower yields the same is rarely noted for methanol even though its capacity to get contaminated is as great or greater. Methanol is best dried by stirring with magnesium or magnesium with some iodine and then distilling.  3A molecular sieves have pore sizes small enough that methanol can compete with water to enter. Nevertheless, it is useful to distill methanol from magnesium and store it over 3A sieves.

Ethanol

The best agent for ethanol is 3A molecular sieve powder. Treatment with sodium followed by addition and refluxing with a high molecular weight ethyl ester is effective because any sodium hydroxide formed reacts irreversibly to form ester carboxylate. Without the ester addition, sodium hydroxide removes protons from methanol to recreate water to some extent.

Charge-Transfer Complexes for Isolation/Separations: Lewis Acid Type

Introductory Base Ideas

The KiloMentor emphasis in process development is on the simplification of the work-up, isolation, and purification of the desired product coming out of a process step.  This is where more of a typical process’s time is expended. Another reason to emphasize this phase is because this is where chemical database searching is poorly organized to help choose the better methodology.

Separations on the molecular level are achieved by phase switches such as liquid-liquid extractions (ie methanol\hexane), acid-base extractions, and physical changes of state (like crystallization, distillation, sublimation, etc). Purification results because the conditions for the phase switch are so controlled that impurities are relatively poorer at making the switch and are thereby left behind.

Charge Transfer Complexes as a Phase Switch

The possible use of preferential formation of charge-transfer complexes in separation and purification is attractive because the property used to achieve the switch is not that of a particular functional group, like the acidity of a carboxylic acid or the basicity of an amine, but is a function of the entire aromatic system and all its substituents. It is more of a molecular property.  Furthermore, because of this, even molecules that have no serviceable functional groups, like naphthalene, for example, can form charge transfer complexes.

Electron-deficient compounds that are known to form pi-type charge-transfer complexes with aromatic compounds include tetracyanoethylene, 3,5,7-trinitroflorenone,  tetranitrofluorenone, dicyanomethylene, trinitrofluorene, tetranitrofluorenone, trinitrobenzene, and bromotrinitrofluorenone. Usually to form a solid thermally stable complex the aromatic system must comprise at least two rings, but tetracyanoethylene forms complexes even with benzene. Newman has reported that the compound 4-bromo-2,5,7-trinitrofluorenone forms a complex with benzene that can be dried under a vacuum of 1-2 mm mercury at room temperature but was decomposed by heat before reaching 100C.

Also, in order to achieve higher stability the substituents must not be ones that can interfere with a close approach to at least one of the faces of an electron-donating ring. As an illustration, a methyl aryl substituent is well tolerated but an ethyl group is not.  This sensitivity to small, what would seem to be inconsequential, variants is part of the attraction to using such a method.

There are very few examples of isolation/purifications using charge-transfer complexes. The only one I could find was a report that a charge-transfer complex was used to remove naphthalene from petroleum distillate. 

No one it seems has succeeded in using one of these electron-deficient complex formers as a component in a reactive distillation to reduce the volatility of one aromatic substance in the presence of another or of an aromatic substrate in the presence of a non-aromatic one. In order to be effective in this situation even a preferential association of one substance over another might be sufficient to be practicable.

Hydroxymethanesulfonic acid as a Volatile Strong Acid Solvent

 

Hydroxymethanesulfonic acid has the unusual property that it can be removed from a mixture by reducing the pressure, whereupon it decomposes to formaldehyde, water, and sulfur dioxide gas which are all easily swept out of a mixture.

Perhaps hydroxymethanesulfonic acid can be used as a solvent replacement for sulfuric acid which is decidedly non-volatile.

 

Because it is still an aqueous acid the solvent leveling effect means it cannot support anything more acidic than the hydronium ion. Nevertheless, it can be employed at high concentrations of these hydronium ions to increase the rate of hydrolysis of functional groups.

It would accelerate the hydrolysis, for example, of sterically

hindered esters which could not easily access a tetrahedral transition state, or be used to convert cyanohydrins to hydroxycarboxylic acids.

  

⍺-Hydroxycarboxylic acid can be a protected ketone or aldehyde that can be deprotected by periodic acid cleavage. In this way, a neutral molecule could be handled as a carboxylic acid up until a final unmasking step.

Hydroymethanesulfonic acid could be usefully used in an acid treatment of any substrate attached to a polymeric resin because it would not need to be removed from the resin by exhaustive washing. Instead, the resin could be filtered and dried under reduced pressure. Hydroymethanesulfonic acid would dissociate and evaporate away.

The green solvent glycerol cannot be freed of sulphuric acid by evaporation of the acid and if it is neutralized the inorganic salts will be trapped in the high boiling solvent. A possible solution is to replace sulphuric acid with hydroxymethanesulfonic acid. This acid can be removed by heating under reduced pressure whereupon it degrades to sulfur dioxide, formaldehyde, and water; all of which can be easily evaporated away.

The volatile strong acids HCl or HBr would react with glycerol irreversibly while there is a good chance that hydroxymethylsulfonic acid would be stable in it.

Hydroxymethanesulfonic acid solution mixed into a higher boiling dipolar aprotic solvent such as dimethylsulfoxide. This mixture might provide greater solubility for organic substrates, a higher reaction temperature, adequate water concentration for reaction, and the capacity to remove both excess water and the acid by heating under reduced pressure.

Hydroxymethanesulfonic acid can be anticipated to be a strong acid of choice when trying acid-catalyzed reactions involving functionalized polymers since it can be removed by placing the resin under a vacuum.   

Hexane- Methanol as a Reaction Medium

 

This idea is speculative. As far as I know, there is no experimental evidence to support it. For example, it is not known for any particular reaction mixture how much of that mixture could be mixed with how much of the hexane-methanol azeotrope and still actually get separation into the two phases at the conclusion of the reaction. Also, even if liquid layer separation is achieved, we cannot know in advance what the difference in partition coefficients between mixture constituents will be even though this will determine how practical the separation will be and how many times the procedure would need to be repeated to get a useful concentration of even one species.

The azeotropic mixture of hexane and methanol split into two phases when cooled below 35℃. At temperatures below the UCST, there exist two phases containing different proportions of hexane and methanol. The azeotrope between hexane and methanol has bp 50℃; its composition is 73.1% hexane and 23.9% methanol by weight. The relative volumes of the upper and lower phases are 67.8: 32.2, about two parts to one part. When the hexane-methanol separates on cooling the composition of the upper layer will be 85% hexane and 15% methanol with specific gravity of 0.675. The composition of the lower layer will be 42% hexane and 58% methanol with specific gravity of 0.724. The difference in densities should accelerate layer separation.

The inputs for a reaction mixture would be combined in the warm single phase above 35℃, with precautions to avoid any runaway exothermicity. Then after the reaction’s completion, it is cooled to split the reactor’s contents into two phases between which the reaction mixture’s substrates would partition. In most cases, the desired product will not be conveniently distributed in only one of these phases. The advantage of the azeotropic mixture is that the reactor contents can be concentrated by distilling the azeotropic composition and then by adding either methanol of hexane as desired azeotrope distillation can be continued to provide the reaction mixture in either hexane or in methanol as subsequent treatments require. It should be noted that it is essential for this processing that the reaction mixture not be degraded during the distillation that is required for the solvent change.

The process is ‘green’ in the respect that distillation should recover a significant portion of the azeotrope solvent mixture since most preferably the azeotrope solution will be concentrated down to the minimum storable volume of the reactor before more methanol of hexane is added to chase the residual azeotrope volume.

Performing chemical reactions in a homogeneous mixture of hexane and methanol have further advantages when one no longer insists upon using the actual azeotropic composition as the composition of the liquid medium. Hexane/methanol mixtures of different relative ratios will span a wider range of solvent environments and this choice can be very beneficial in optimizing reaction conditions. A variable parameter such as a ratio of solvents is much easier to work with than an invariant parameter.

A mixture of solvents provides the opportunity to dissolve some reactant first in one of the pure solvent components and then dilute with the second pure solvent component to reach the optimized ratio before commencing a reaction.

These substrate concentrates could then each separately again be dissolved in a new portion of the single phase azeotropic above the 35 C UCST and again cooled and the phases separated and evaporated. These oily residues are treated just the way solutions are treated in a liquid-liquid extraction to improve the degree of separation of the substrates.

The disadvantage of the methodology will be that each time the extracts need to be evaporated essentially to dryness because otherwise the proportions of liquids hexane and methanol will change so that the two phases will not continue to separate.

The advantage is that because both phases contain the same two solvents-just in different proportions, the partitioning of substrates between the phases might be expected to be closer to 1:1 and so the selectivity in the partitioning of two similar substrates might be more sensitive. That is to say, one substrate might be slightly more soluble in the methanol-rich phase while the other might be more soluble in the hexane-rich one. 

If no separation of liquid phases occurs at any stage it is possible that a larger proportion of the hot azeotropic hexane/methanol mixture is required. This is easily rectified. Simply add more of the hot azeotrope composition and rewarm the total solution until above the UCST and recool.

Improving Throughput in a Chemical Reactor of Fixed Dimensions

In a multi-purpose chemical plant, there are only a fixed number of chemical vessels, each one with a fixed volume capacity, and each with its own fixed auxiliary facilities ( ie. stirrer types, materials of construction, distillation facility,  heating, cooling). More of a chemical product can be produced in a reactor, whatever its type, by increasing the concentrations of reactants and reagents in the vessel; but, there are limitations to the extent to which this can be done.

Improving the throughput is not equally important for all the process steps in a particular synthetic sequence. It is nearly without exception that it is the earliest steps in a synthesis that turn out to need to be repeated multiple times to provide sufficient early intermediates to make the desired quantity of the ultimate desired product. It is reducing the number of repetitions of these early steps that can deliver cost savings.

Improving the throughput of a process step is frequently the work of a chemical engineer rather than a process chemist. Nevertheless, a chemist may be assigned such work or the scientist who developed the laboratory procedure upon which the process step is based may be able to provide valuable insight or even experimental assistance to speed up or improve the change.

The wish to improve the throughput of an early step in a process sequence comes after the process itself has been successful at scale. Therefore, changes that might improve the throughput and decrease the number of times an early step needs to be repeated absolutely must not involve reengineering the whole process. The changes need to be ones that can be expected to have essentially no impact on the quality or quantity of the output of the step being modified. For example, completely switching to a different solvent for the reaction would almost certainly be too much change.

Very often a reaction has been first scouted and subsequently improved in the laboratory under conditions where a reaction occurs under homogeneous conditions. Enough solvent is specified to completely dissolve all of the reactants and that amount of solvent sets a minimum volume constraint for any given amount of reactant(s). This limitation may not be real. So long as other constraints do not intervene, experimentation may be able to demonstrate that a reactant does not need to be completely homogeneously dissolved in solvent throughout all or some interval of the reaction time— but this needs to be established by appropriate laboratory experiments using the actual qualities of reactants and reagents that will be used in the plant or pilot plant. This increases the number of experimental runs— never a desirable thing— and requires that materials representative of what will be used at scale be already available to the lab. This is important because the physical properties of undissolved solids can change reaction rates substantially.

The most frequent impediment to increasing the concentration of reactants, reagents, (and catalysts) to increase the throughput and reduce the overall process cost is the exothermicity of the reactions involved. As the reacting species are brought more intimately into contact they react faster and more heat can be produced. The more exothermic a transformation the more severe this restriction becomes. Almost always there is a temperature range that must not be exceeded.

The primary function of a reaction solvent is to shape the physical environment under which the reactants meet. A very prominent secondary function is to provide the capacity to absorb and buffer the energy absorbed or released by the reaction. 

If we want to increase the concentrations of the reactants in our reactor so that we can increase the batch size in our reactor this will produce a larger and more rapid exotherm. To keep the reaction temperature within a specified range therefore we need to remove heat from the reactor’s contents more quickly.

The most common way to do this without changing the reactor or its cooling facility is to periodically stop further reaction, cool the reactor contents to the bottom of the reaction’s acceptable range, and then cause the reaction to resume until the internal reaction temperature reaches the top of the acceptable range; then stop the reacting and repeat. This starting and stopping of the reacting process is most commonly done by starting or stopping the gradual addition of one of the reactants. An even more common variant is to adjust the addition so that exotherm and cooling capacity remain balanced within the acceptable range.

To an extent to be determined experimentally for any particular transformation higher throughput may be achieved at the expense of a longer addition time for reactant mixing together. As we shall see other changes can be blended in with these two.

For most general-purpose chemical reactors cooling is provided by an external cooling jacket, a liquid coolant flowing in that jacket, and external refrigeration capacity to recool the refrigerant. The cooling system has, within its own limits, an adjustable capacity to keep the portion of the reactor wall that is in contact with the reaction mixture cold. The effectiveness of cooling is not however a function of the refrigeration power alone. Cooling of the bulk of the reacting mixture will depend upon the effectiveness with which the bulk reaction mixture is brought into close contact with the cold refrigerated reactor wall. This will depend upon the quality of the stirring in the reactor itself. The stirrer in a given reactor is a given for our purposes. Its velocity has an upper limit and its mixing effectiveness will depend upon other things, most prominently the viscosity of the reaction mixture, which itself changes through the reaction period. It also depends upon the ratio of reactor volume/wall surface area.

It is an impending difficulty that in the plant the reactor volume/wall surface areas are much higher than in typical laboratory reactors. Cooling the plant can be much less effective because of this. On the other hand, stirring and refrigerating power may be substantially better in the plant than in the laboratory. I think it is going to be impossible to model in the lab.

What can be done is a very low-risk experiment at plant scale in which an acceptable increase in reactant/solvent ratio is applied combined with maximum cooling power and maximum stirring speed while controlling the rate of addition of the limiting reactant that will hold the reaction temperature within the allowable range. This experiment will provide you with the minimum addition time possible for the throughput you are trying. 

Every solvent has its own heat capacity. Heat capacity is the number of calories per mole required to raise its temperature by 1 centigrade degree. Different solvents will have varying abilities to absorb heat but we have already agreed that changing the solvent for a reaction would be too great a change to countenance at this stage when a process has already been proven acceptable. Could we add some new solvent into the reaction mixture to produce some solvent blend with a higher heat capacity? This is not likely to work well. The addition required to substantially change heat capacity would probably be sufficient to substantially change the reaction conditions.

The heat capacity of the solvent is not the only means by which the exothermic energy of chemical change can be controlled. Every solvent has a particular heat of vaporization whereby calories are removed to boil that solvent. If a solvent mixture is used as a reaction mixture at the temperature that is its boiling point heat from the reaction mixture can be removed by using it to boil solvent up into an attached condenser where it is cooled and that liquid will flow back into the reactor.  This source of cooling can supplement cooling provided by a cooling jacket.

The downside of this technique is that it only operates when the reaction temperature is the same as the boiling point of the solvent. The optimal solvent range is usually chosen to maximize the selectivity of the reaction not to match the solvent’s boiling point so this method is unlikely to be applicable for increasing the throughput of a process already optimized.

However, a variant of the method might be useful. Suppose we were to modify the reaction mixture by adding a small amount of a cosolvent that had a boiling point within the reaction’s optimal reaction temperature range. So long as this cosolvent does not form an azeotrope with the other reactor components, it will boil at its own boiling point, take up heat from the reaction mixture, and vaporize up into any attached condenser. In the condenser, the heat will be extracted and the condensed cosolvent returned to the reactor where it can repeatedly be revaporized removing heat each time this occurs. This removal of heat will modulate the reaction exotherm and add to the cooling capacity. After the reaction period, the cosolvent is evaporated off and not returned. The reaction mixture is unchanged! Because a small amount of cosolvent is evaporated, and condensed over and over again a small amount of cosolvent could provide a lot of cooling.

The cosolvent for this use may be many things; all that is required is that it be unreactive under the reaction conditions and that it has a satisfactory boiling point and heat of vaporization. The condenser’s cooling capacity must also be sufficient. This cosolvent need not be a pure substance. It could be a mixture of liquids of the correct composition to have a lower boiling azeotrope. Low boiling examples could be ethyl ether/isoprene bp. 33.2 C or ethyl ether/ methyl formate bp.28.2 C. It could even be condensed, precooled gas that is sparged into the fluid reactor contents and then vents without condensation; liquid nitrogen for example.

Some solvents can be partially frozen onto the reactor walls before beginning the addition of a reactant that starts the reaction process. Using this technique the heat of melting of that solidified solvent can be used to cancel the calories produced by the reaction. This modulation would very efficiently cancel the initial burst of heat from the reaction before the cooling from the cooling jacket kicked in. Only a limited number of solvents have freezing points and heats of fusion in useful ranges. Water, DMSO, glacial acetic acid, dichloroacetic acid, dioxane, ethylene glycol, formic acid, formamide, nitrobenzene, and glycerol. The melting point must be low enough and the heat of fusion high enough to be useful.

The limitation on throughput using a particular reactor is often the point of maximum volume in the procedure. This point of maximum volume often comes at the reaction quenching or extraction stage. In the strict sense, these points of maximum volume are for the process step not strictly speaking for the reaction period itself.

Let us look at the situation that arises where the point of maximum volume comes after the reaction mixture has been quenched with an equal volume of water to provide a two-phase mixture. It may be possible to transfer half of a reaction mixture into a second vessel (it need not necessarily have the same facilities that were required during the reaction phase proper ie. heat/cooling facilities; it may be no more than a stirred tank ) and then quench each portion separately in its own vessel and work up each vessel’s contents separately, thereby doubling the throughput of the reaction. In the same manner, a process step that has its point of maximum volume during a liquid-liquid extraction can have the reaction mixture divided between vessels and the extractions done separately each in its own vessel. The extracts can then perhaps even be recombined and the remainder of the isolation/purification done together.

Mixtures of Triethylamine and Glacial Acetic Acid for Extraction or Extractive Distillation

Triethylamine is an organic liquid. Glacial acetic acid is also such a compound. Both are sufficiently inexpensive that they could be used as solvents for organic syntheses. The former is distinctly basic and the latter acidic. 

Triethylamine/acetic acid forms a constant high boiling azeotrope bp. 163.0℃ more than 40 Cº above the acetic acid. The azeotropic composition is 69% glacial acetic acid and 31.0% triethylamine by weight. This is a molecular ratio of 3.75: 1.0 acetic acid: triethylamine or 14: 4 in nonfractional units. If we imagine a medium in which the triethylamine and acetic acid molecules neutralize each other in a 1: 1 ratio, this azeotropic composition will be distinctly acidic because it contains a substantial excess of the carboxylic acid. Nevertheless, the excess acetic acid cannot be distilled away. It forms a constant boiling mixture. The upper reaction temperature for this mixture would be its constant boiling azeotrope temperature of 163.0℃.

It is this azeotrope that is a principle subject of US 3,244,761 where at column 9 starting at line 71, it is remarked, “It should be emphasized that even though molar excesses of acids are normally used in the preparation of my selective solvents (of which the triethylamine/acetic acid azeotrope is one) the corrosively of the solvents towards materials vulnerable to acid attack is unexpectedly low, sometimes approaching that of distilled water….The reason for the surprising lack, or minimization, of corrosively in my selective solvents is not understood with certainty.”

More will be related from this patent later.

A different mixture of these two substances has also been described. It is reported that a 1: 1 molar ratio of the two components forms something resembling an ionic liquid. Such a composite would comprise 101 gm of triethylamine with 60 gm of glacial acetic acid. This is 62.73% triethylamine and 37.27% acetic acid; almost the exact opposite of the azeotropic composition. This mixture would be expected to be basic. Upon boiling about three-quarters of its triethylamine would be expected to distill off at its bp of 89.5℃ leaving a composition of the azeotropic mixture.  

Both of these aforementioned mixtures may be expected to show useful, and even unusual solvent properties. Both glacial acetic acid and triethylamine as pure fluid compositions, on their own, can be expected to dissolve organic compounds more readily than water. As mixtures with either strongly hydrogen-bonded species or solvent-separated ion pairs (1 : 1 combination of acid and base), an even more comprehensive range may be soluble. Also, because the medium will cause the components to strongly associate with each other, the dielectric strength may permit various inorganic salts also to dissolve appreciably. 

We also know what happens when water is added to one of these compositions. The main subject of US 3,244,761 is the use of triethylamine/acetic acid high boiling azeotrope, among other acid-base combinations, for the solvent extraction of polycyclic aromatics from crude hydrocarbon streams. Therein, it is reported that dilution with water causes the triethylamine and the acetic acid to be dissolved in the bulk water enabling any more hydrophilic elements to be recovered. It also reports that mixtures of high boiling azeotrope of triethylamine/acetic acid can be recovered from a mixture with water simply by distillation whereupon the water comes over first at 100℃ and then the azeotrope at 163℃. The evidence for this is in Example VIII of this patent.

This patent also teaches “The solvents of this invention (of which triethylamine/acetic acid is one) in addition to being highly selective towards aromatic hydrocarbons, exhibit high solvent power..”

At another point, ‘still another way in which my process (of solvent extraction) can be carried out is to employ an anti-solvent in conjunction with the amine-acid solvent…The use of such anti-solvents in hydrocarbon extraction processes is well known….[t]ypical anti-solvents for purposes of this invention are kinds of paraffin, such as pentane, heptane, octane, isooctane…”

This latter makes evident that triethylamine/acetic acid is not significantly miscible with such saturated hydrocarbon liquids.

N.B. It is noteworthy that US 3,244,761 also identifies the high boiling azeotrope of triethylamine and acetic acid as a useful ‘separation solvent’ for use in extractive distillation. It could be used for separating compounds neither of which boils above about 160℃ (the bp of the azeotrope is 163℃ ) and where one of the components has an aromatic substructure and the other does not. The more aromatic material will be held back in the still pot. If a composition with a higher upper-temperature limit is required US 3,244,761 suggests a mixture of tributylamine and caproic acid or a combination of dimethylaniline and benzoic acid. Both these mixtures would have bps greater than 450℃.

Isopropyl Alcohol/ Toluene/ Water: A Thermomorphic Ternary Azeotrope for Extractive Purification

 

There are several references to this Ternary System: Isopropyl Alcohol, Toluene, and Water at 25°C.

E. Roger Washburn and Albert E. Beguin, J. Am. Chem. Soc. 1940, 62, 3, 579–581; J. Am. Chem. Soc. 61,1694 (1939); 54, 4217 (1932); 56, 361 (1934)

It is well established that isopropyl alcohol, toluene, and water mixtures upon distillation pass over into the distillate as a tertiary azeotrope that boils at 76.3 C and has the weight composition 38.2% IPA, 48.7% toluene and 13.1% water. When it separates into two liquid phases the upper layer is 92% by volume and the lower 8% by volume. The composition of the upper layer is 38.2% IPA, 53.3% toluene and 8.5% water and has specific gravity 0.845. By contrast the small lower layers 38.0% IPA1.0% toluene and 61.0% water has specific gravity 0.930. This behaviour of splitting into two layers on cooling is called thermomorphic.

What is not apparent from my examination of what literature I can access is the temperature at which these two phases merge into a single phase. That is: What is this mixture’s upper critical solution temperature (UCST)?

To be useful in the application I am contemplating there needs to be a manageable difference between the lowest temperature needed to get a homogeneous solution and the azeotropic boiling temperature. That is, specifically the UCST needs to be less than say 50 C.

My idea is to use portions of the lower phase of the separated ternary azeotropic composition (with a composition of 38.0% isopropanol, 1.0% toluene and 61.0% water)  in the volume proportion of 8 volume % versus the 92 volume % of the upper phase to sequentially wash the upper toluene rich phase, so as to remove somewhat preferentially more polar constituents from a mixture of substrates initially dissolved in the homogeneous single phase ternary azeotropic combination of isopropyl, toluene and water.

This, it is hoped, will produce a result similarly to what is called ‘swish’ trituration. In ‘swish’ trituration a solid mixture of substrates is repeatedly triturated with an anti-solvent in which the predominant component is very nearly completely insoluble but in which the impurities are meagerly soluble. Even so, they are removed because of the substantial quantities of the triturant used. If this were to work, the result would be the purification of the main component.

This is how I imagine the process would be executed.

Experimental

10 liters of the IPA, toluene, water tertiary azeotrope are prepared. The mixture is heated to a temperature conveniently above the UCST and divided into two portions, one of about 1 liter and the second about 9 liters. The 9 latter portion is allowed to cool below its UCST and the lower more polar phase separated and stored in a stoppered vessel. This phase labelled (A) will provide the multiple wash portions used to remove more-polar components of the mixture of substrates. In a separatory funnel, part of the 1 liter portion of the still warm, still a single phase tertiary azeotrope which we call (B) is used to dissolve the mixture of substrates to be separated. For about 2 grams of mixture, 100 ml of warm homogenous azeotrope is used. The test mixture must dissolve completely at the warm temperature where there is just one liquid phase and when the solution cools it is necessary that two liquid phases separate. 

It is essential that two liquid phases form even though the presence of the charge of substrates is included. It is for this reason that I am choosing to only use 2% weight to volume (substrates to solvent). If two phases do not separate, it will be necessary to increase the amount of azeotrope solvent mixture until they do. On the other hand, for reasons of the throughput of the purification, it is desirable to use as large an amount of substrate mixture as is consistent with retaining two separating liquid layers.

When the liquid in the separatory funnel has cooled to room temperature, remove the smaller volume of the lower phase into a graduated cylinder and note the volume of this phase. Transfer this, what I will call the 1st wash, to an erlenmeyer flask and save it for analysis to learn the degree to which you have concentrated the polar impurities in this 1st wash.

Now, into the separatory funnel add the exact same volume of  (A) as you have removed in the first cut. Two phases should be present since what you are adding is pretty close in solvent composition as what has been removed. Warming in the separatory funnel to above the UCST will produce one phase and cooling back will split the volume into two again. The substrates, which we seek to separate, will have again partitioned between the layers. Cut again and move that layer into a second erlenmeyer flask for analysis.

How successful the technique is for purifying a major less-polar substance will depend upon the substance being substantially more soluble in the toluene-rich layer than in the water-rich layer. The more polar impurities need not be particularly soluble in the water-rich phase so long as they are more soluble than the major component. Poorer solubility of the polar impurities only means a larger number of these small volume washes will be required. Note that the size of the washes must be the same as the volume of the first lower layer. If not the composition of the ternary azeotrope in the separators funnel will change too much and the layers may no longer separate properly. 

An advantage of working at-scale in a closed inverted reactor as the separatory vessel is that a separation temperature as low as -20 C can be used because extraneous water from the plant air cannot be condensed into the liquid medium from the air— the liquid layers are covered by inert gas!