Advanced Manufacturing Ideas Can Be Applied to Fine Chemicals/ Pharmaceuticals

 

In their August 12th, 2023 issue pg. 63-64, the Economist magazine describes the advanced manufacturing of first, computer chips and then, cordless electric drills. Reading this brief report suggested to KiloMentor possible parallels with future advances in the scale-up for the manufacturing of complex organic chemicals.

The article points out that “chips are designed using software that directly links to the automated hardware which fabricates them.” The efficiencies that this unlocks derive from the consequence that “the constraints of the production line- even fiddly details like the positioning of screws— are encoded in their CAD (computer-aided design) programs.”

Well, how does this have any analog implications for chemical processes for making sophisticated chemicals? We don’t have automated production facilities and we don’t work out the details of process steps in computer programs.

The research laboratory functions as our design tool and the pilot plant functions as our automated fabrication hardware and our problem is too often that our designing is not sufficiently linked to our manufacturing. The problem is how do “even [these] fiddly details”, like the constraints of large-scale production get signalled back to the laboratory before valuable time is wasted?

In advanced manufacturing, these constraints are wired into computer-aided design programs. For our projects, there are two possibilities. Either our process design chemists must have these limitations wired into their chemical know-how or the whole company must adopt some form of what has been termed ‘full process vision’. I have written previously about this idea in the blog, Avoiding the Screw-up from Left Field with a Full Process Vision.

Since it is difficult to impossible to find this article I have reprinted it below.

"Some process chemists will find themselves as small cogs in large teams whose goal is to develop new specialty chemicals or pharmaceuticals.  As a scientist whose contribution is to apply highly specialized knowledge, you may be bunkered in a rather isolated trench or silo within your organization. Your mission may be defined for you rather narrowly so your undoing may come from an irrefragable requirement that comes from outside your silo and that is imposed so late in your work plan that it really means starting over.    

A powerful organizing structure for pharmaceutical product development is presented in an article by Pradir K. Basu, Ronald A Mack, and Jonathan M. Vinson, “Consider a New Approach to Pharmaceutical Process Development“ Chem. Eng. Prog., 95(8), 82 (1999).  It seems intended to reduce the likelihood of the above misfortunes.  

Process chemists, as knowledge managers, need to press at an early stage in their work for some mechanism within the wider team so that these must-have ‘requests’ from outside your core group reach you before your work is too far advanced. 

Much of the referenced article presents no more than standard reminders of the importance of cost considerations throughout discovering a synthetic method, scaling it up, and putting it into production for a process to manufacture a new pharmaceutical. This is the pharmaceutical business with the marketing, selling, and regulatory functions stripped away. Its importance to corporate profitability does not engender much debate. The importance of the article is that their concern is broader. 

The authors are concerned about the efficient execution of a plan that starts after identifying a candidate to be a commercial drug with a salutary effect on a biological target and proceeds to the validation of manufacture for that molecule at a commercial scale. 

The enhanced approach that they propose identifies what they call ‘process vision’ as the core organizing principle. The definition and exemplification of the expanded concept of ‘process vision’ is the article’s significant accomplishment. 

The authors help us understand different aspects of this 'process vision' at different points in the article. For me, I cannot say I adequately grasped what they were getting at until I drew particular phrases together from my notes. Some of these quotes, drawn from different parts of the essay are: 

 “The process vision satisfies all essential requirements, including those for safety, quality, waste minimization, cost, time, and operability”. 

“The process vision is neither the process with maximum yield nor the one that gives maximum product purity…..it is neither a chemist’s vision nor an engineer’s vision; it is not even the vision of the chemists and engineers together.” 

“It is a vision that all stakeholders in development, manufacturing, and marketing can share…..” 

Reading between the lines and amplifying certain aspects, the process vision emerged as a policy statement that provided, as a starting point, standards by which team members coming from each stage of the organization's endeavor (laboratory process, kilo lab, pilot plant, and manufacturing facility) could satisfy downstream colleagues’ concerns from the outset of their own work. The authors' specific examples of the unique orientation and emphasis that players at the different stages have and which they want to see addressed from the very outset reinforce my interpretation. 

This early overview, whose importance they emphasize, can be expected to show up inevitable cross purposes and improve the odds for early compromise and conflict resolution. 

They write:

 “Chemists think in terms of steps, reactions, yield, purity, and so on; engineers in terms of unit operations, physical properties, heat load, and the like; manufacturing personnel in terms of throughput, waste control issues, and plant modifications that may be required to run a process; and marketing people in terms of the net present value of the product, how much it can sell for, etc.” 

“It is important ….to get stakeholders to develop….agreed-upon objectives of process development.” 

“communication among….personnel is critical during process development.” 

“We need to…. provid[e] development team members with systems or tools to facilitate communications among different disciplines.”

“Unless the manufacturing team is involved in the process development, they will not have confidence in the scale-up”. 

“…manufacturing and commercial input at this stage [late stage discovery] are essential for choosing the optimum processing route”. 

“Team members need to be involved in setting targets for cost, manufacturability, waste and emission loads, development time….” 

“These alternatives must be evaluated based on….criteria agreed upon by all stakeholders….” 

“If stakeholders are involved in planning experiments, it’s likely that more useful data could be collected from fewer experiments.” 

For me, the management tool the authors recommend for achieving this widely held ‘process vision' is Panglossian. 

The authors propose that even at the experimental program level one should try to bring together a diverse project team including representatives all the way out to marketing, frequently enough to work out priorities and make decisions. This is what they recommend. 

This seems excessively optimistic as regards human nature. Instead, I suggest, one could establish a 'process vision' statement establishing some sort of median or normal starting-point performance criteria that would address recurring diverse concerns of process development, manufacturing, regulatory affairs, and marketing and that would chevvy the most common interests of the downstream project teams on the upstream collaborators. In this implementation, the process vision would be via a statement delivered with full corporate authority that would continuously challenge upstream groups with the standard core concerns of the downstream members. 

The authors illustrate marvelously this challenging interaction throughout their article. What I interpret them to be saying is that the problem is not that different elements of the project team have concerns that inevitably seem to operate at cross purposes; but that the team members can reach solutions that satisfy all parties, so long as the areas of tension are discovered early enough. 

KiloMentor has a strong preference for its alternative. The use of a process vision statement as a proxy for the perspectives and concerns of downstream project groups seems preferable to using large frequent group meetings to actually direct even the collection of particular data. For a company’s drug product projects to be successful and on time, any process’s strategy must not conflict too greatly with the psychological needs and private professional goals of the individual team members. The people downstream in the project, whether they be in late-stage process development, manufacturing, or marketing, simply will not give a project the attention it needs until it arrives at the phase where they are being held singly and personally responsible. They are too busy concentrating their attention on what is on their plate already and extinguishing the fat that is already on fire. This is human nature! Besides, pharmaceutical product projects can go on so long that some participants can realistically expect to no longer be involved when a late-stage discovery project limps into manufacturing or marketing. People may hope or plan to outrun the difficulties. Only unambiguous corporate endorsement can get everyone to give a thought to early-stage projects.

Equally problematically, the upstream professionals, working at a particular phase of the work on their own turf, would require an uncommon personal modestly to accept without rancor face-to-face demands that particular questions be answered on a priority basis. 

A corporate ‘process vision’ statement takes the personalities and egos out. At the same time, the standards proposed by a process vision statement would command authority and yet not be carved in stone. They would exist to bring a persistent awareness of particular concerns. They would bring those different needs, which may be pulling at cross purposes to early attention, and they can be expected to bring the affected team members together as needed to create or negotiate a solution." 

Solvents and Additives that Qualitatively Modify the Comparative Reactivity of Hydride Reducing Agents

Sodium borohydride, lithium aluminum hydride, and sodium bis(methoxyethoxy)aluminum hydride (Vitride) are the most common hydride-reducing agents available for organic chemistry. These super hydrides can be used where high stereoselectivity is required.

What is comparatively less recognized is that both the reducing power and the selectivity of reducing agents can be dramatically switched depending on the solvent in which the reduction is performed. So much so that this may be an easier means to achieve a particular selectivity than changing the reagent itself.

S. Krishnamurthy published a note Solvent Effects on the Relative Reactivity of the Functional Groups Halide and Tosylate towards Lithium Aluminum Hydride in J. Org. Chem. 1980, 45, 2550-2551 teaching the dramatic effect of solvent choice on the competitive reduction of alkyl halides (chloride, bromide, and iodide followed a similar pattern) in competition with the tosylates of the same hydrocarbon substructure.

As one would predict, for any ethereal solvent the reduction with lithium aluminum hydride followed the rate Cl<Br<I, and the order of increasing rate with varying solvent was diglyme  >monoglyme >THF >ethyl ether.

What was startling to me was that for tosylate the rate of reduction was about the same as bromide but the selectivity of reaction velocity according to solvent was quite the opposite: ethyl ether >THF> monoglyme >diglyme

The practical significance of the difference turns out to be that in a competitive reaction between a tosylate and a halide in the same molecule, one can achieve selective reductive of one or the other group simply by changing the solvent.  In diethyl ether, the tosylate will be selectively reduced to hydrocarbon in the presence of the halide while in diglyme the halide will be selectively removed in the presence of the tosylate.

Dr. Krishnamurthy explains this using the theory of tight and solvent-separated ion pairs.  He proposes that in diethyl ether the lithium cation is poorly solvated and forms a tight ion pair with the hydride donor which is less nucleophilic towards the halides. But because the tosylate group has its own oxygens these preferentially complex the lithium ion and increase the reactivity of the tetrahydridoaluminate which attacks by a pseudo cyclic transition state with a reactivity more like a solvent-separated anion.

When the solvent becomes increasingly able to solvate the lithium-ion the result is a solvent-separated hydrido aluminate anion whose reactive order is iodide> bromide> tosylate >chloride.

The KiloMentor blog has discussed means to exchange such high boiling solvents as diglyme or monoglyme, making them more practically useful as solvents.  It would also be interesting to explore the relative reactivities of halides and tosylates when treated with lithium aluminum hydride in the presence of say bis-trimethylsilylated polyethylene glycol because this polymer can be removed from a reaction mixture by precipitation with either diethyl ether or hexane.

Effect of Lanthanides on the relative reactivity of Carbonyls (aldehydes versus ketones) to reduction

Aldehydes are more easily reduced than other carbonyls, right?

Well usually, but the selectivity can be reversed dramatically and inexpensively.  In a communication in 1979, Jean-Luis Luche and Andre Luis Gemal published a communication, J. Am. Chem. Soc. 101(19) 5848 1979, teaching that in reductions by sodium borohydride in ethanol/water of mixtures of carbonyls, or different carbonyls in the same molecule, the addition of one equivalent of a transition metal salt so effectively catalyzed the formation of geminal diol in the case of saturated aldehyde that a ketone would be reduced preferentially.  The preferred transition salt was cerium chloride. A 50% excess of sodium borohydride was needed because the metal accelerated the reaction of hydride with solvent as well.

This could be a very desirable outcome.  It would be useful to be able to mask a ketone by reduction and then use an aldehyde in the same molecule as a reactant in some other transformation where it is reactive but in the presence of the ketone would produce unwanted by-products.

Filtering and Washing Organic Crystalline Solids

 

The filtration rate for recovering organic solids from organic liquids in the laboratory is not normally a concern. However, in the plant, slow filtration can extend an operation by several shifts. Reduced throughput in the plant costs significant money. Increasing the throughput in the laboratory can be just a matter of putting more reactants in a larger reaction flask and a larger bulk of crystals into a larger filter. Throughput in the plant is more often increased only by reducing the cycle time; one's choice of reactors and filters is small or nonexistent. Still, slow filtrations increase cycle time.

In the laboratory, the norm is water pump-vacuum filtration with a paper filter medium. However, paper filters clog more readily than filter pads. In process development laboratories filtration testing should use the identical filter pad material that is planned for usage in the plant. The material needs to be cut to size to fit the laboratory filter funnel which is usually of the Buchner variety.

Most preferably the size of the Buchner filter funnel to do the laboratory testing should be chosen so that the final cake thickness at the plant and laboratory scales are similar. If the laboratory filter cake is thicker the prediction will be even more conservative and hence dependable.

Normally the rate at which the product slurry is poured onto the filter is slower in the plant than in the laboratory. 

Filtering duration in the plant is longer than in the lab because the filter cake is usually thicker.

In this discussion, I will need to make clear the boffin’s argot for this technical area. The liquid accumulating after passing through the filter is called the filtrate. The solid collected on the filer pad is called the filtrand. ’Washing’ means cleaning a filter cake. The predominant contaminants being removed are product impurities that have not precipitated out of the crystallizing solvent. This ‘washing’ is distinguished from ‘cleaning’ which refers to cleaning parts of the filter equipment itself,. Washing the equipment to remove adhering solid is instead called ‘rinsing’ (e.g., rinsing a filter screen or a filter cloth with jets of water). 

A distinction is made between washing and extraction or leaching. ‘Washing ’ means the elimination of a solution of both dissolved contaminants and residual dissolved target substance (the mother liquor) from the pores between the particles/crystals of a filter cake. ‘Extraction’ or ‘leaching’ is a subtype of ‘washing’ that additionally draws soluble matter out of the interstices of the solid particles themselves. 

The word ‘drying’ refers to thermal drying wherein volatile materials, predominantly solvent or a processing liquid, are removed by the application of heat and sometimes lower pressure. The reduction of liquid from the filter cake by mechanical means is called ‘deliquoring.’ In the laboratory with the standard Buchner filtration, deliquoring occurs during that brief moment between when the final portion of slurry liquid disappears into the surface of the filtrand and when, with a rushing sound, a flurry of bubbles exit the throat of the porcelain or glass filter assembly.

The operation called ‘expression’ is the separation of liquid from solid-liquid systems by increasing the pressure on the material. The liquid is pressed out from the solid.

Filtrations differ in where the filtrand particles are retained. If the particles are generally large compared to the pore sizes of the filter pad, the particles will predominate on the outer surface of the filter pad and will end up in the cake supported on the filter. This ‘cake filtration’ is most common for recovering organic crystals from recrystallizations. Filter pads very often outperform filter papers. Filter papers more frequently clog because too many small particles get trapped in the channels within the thin paper layer.

‘Depth filtration’ on the other hand, occurs when the solid is trapped inside part of the filter medium. This arises, for example, when the organic solid is filtered in admixture with a filter aid. The completely insoluble filter aid particles provide an additional surface beside the filter pad for the organic crystals to agglomerate on. This speeds the separation from mother liquors but adds a further necessary step separating the desired organic solid from the filter aid. Using a filter aid to speed up filtration can be helpful particularly if it does not significantly add time to later processing steps. The use of filter aids will be treated in a different blog article.

The pressure difference across a vacuum filter system is very limited, and the residual moisture in its filter cake is higher than with pressure filters. Pressure filters allow high-pressure differences from top to bottom of the separated cake. They are preferred when the product must be kept in a closed system for safety reasons, or if the residual moisture content is important. Access for the handling of the filter cake is obviously more difficult using a pressure filter. 

Filtration by centrifugal force, which is another possibility, has the shortcoming of needing more complex capital equipment, but as a general rule, it yields solids with lower residual moisture.

Because most filtrations of crystals in organic chemical synthesis are vacuum filtrations, the pressure drop is never more than 1 atmosphere. 'Expression' from the filter cake is limited and in fact, 'deliquoring' is normally no more than a very brief passage of air or inert gas through the cake. Washing normally quickly follows this draining with gas assistance. The passage of gas through the cake can quickly start some drying action, with concomitant deposition of impurities from any residual mother liquors still on the surface of the product crystals.

Washing can start immediately after the first gas passes through the cake signaling the end of deliquoring or it can start even before any gas enters the cake. The more volatile the solvent the more quickly washing should follow filtration. Washing is typically done using solvent as cold or even colder than the slurry that was filtered. The washing liquid is usually first used to clean the vessel from which the slurry came. A major concern during washing is losing, by dissolution, any of the desired product already trapped as solid and lying on the filter pad.

Wash liquid flows through the filter cake and displaces mother liquor, a mixture of both liquids leaves the cake as a wash filtrate. If the wash liquid is collected separately from the filtrate, analysis can give a good indication of its efficacy. 

The crystal bed of some substances shrinks horizontally and cracks if the solvent is completely sucked away. In such a case, it is particularly important to add the wash liquid to the crystal bed’s surface before this has a chance to happen. Whenever the crystal bed cracks, the washing will become very inefficient, and a large amount of mother liquors will likely be retained in the crystal mass and end up getting dried with the bulk of the solid product; trapping impurities. When cracking is characteristic of the filtrand, the wash should be started even before all the mother liquors have passed into the surface of the cake bed and before any deliquoring.

If shrinking during washing is a washing problem, slurrying the cake in wash liquid and filtering it again is indicated. 

An acceptably small amount of washing will not remove all the mother liquors. Plug flow of the wash liquid cannot be expected through a filter cake. A filter cake contains stagnant zones or dead-end pores, which are not reached by the flow. They deliver their contaminant content by diffusion to the effluent if they do it at all. Diffusion is an asymptotic process and depends on the time elapsed, not on the quantity of liquid. Wash liquid threads its way through the filter cake in what is called ‘fingering.’ In most washing processes the wash liquid has a lower viscosity than the mother liquor. It tends to flow through the cake in finger-like streams past islets of more viscous fluid. This is a random (stochastic) process,  and the local concentration in the cake and the momentary concentration in the effluent will vary randomly. Scale-up from a laboratory test filter even with only small-scale fingering can be misleading as can the results of a few small samples taken from a big cake. 

Sometimes relieving the vacuum while the wash liquid is still in the cake will provide more time for impurities to diffuse out of dead zones into the bulk wash liquid so that they are removed when vacuum pressure is reapplied. The downside of this modification is that it allows the wash liquid to warm somewhat, possibly increasing the loss of the desired product.

In the laboratory, when the washing is over, air or inert gas is typically drawn through the cake for some period. Also sometimes, in the laboratory, some additional wash liquid may be expressed either by pressing down on the cake’s surface with a flat, inert tool or more efficiently using a section of dental dam that completely blocks the entry of gas into the cake.

In the plant, the surface of the cake is rarely manipulated. The only exception is when cracks arise in the cake in which case they must be closed mechanically before continuing.

Drying at scale in large drying ovens adds appreciably to the processing time for many products. When a high boiling solvent is involved, it may be cost-effective to add an additional wash with a different solvent in which the desired product is very poorly soluble but which itself is much more volatile.

This can significantly shorten the overall drying time.

Common Organic Functional Group Derivatives that Facilitate Work-up, Separation, and Purification

 

The most common functional groups in organic molecules all have some reversible derivatives that exhibit acidic, basic, or heavy metal complexing properties.

• The alcohol group can be converted into an O-sulfate that is acidic. 
• The ketone group can be converted into an oxime that is acidic. 
• The carboxylic acid ester can be converted into an acyl hydrazide which is mildly basic and forms stable isolable complexes with heavy metals.

When these common functionalities are present in a reaction starting material but the planned transformation is occurring elsewhere in the molecule replacing alcohol, ketone, or ester group respectively with O-sulfate, oxime or acylhydrazide often could make no difference at the reaction site but could facilitate the isolation of product from the subsequent reaction mixture. 

Perhaps the reason this is not done is because the potential interference of these derived functionalities is not so well understood while for the parent functions there are usually many precedents. Another reason could be that the ease of work-up, separation, and purification is simply not considered sufficiently important to warrant special facilitation.

Testing Catalysts and Inhibitors to Change the Ratio of a Desired Product and an Undesired Coproduct

Suppose you are working with a reaction step that is providing, rather cleanly, a mixture of two products-- the desired one and an undesired byproduct. Perhaps modifications of conditions are not beneficially changing this product ratio. Perhaps a change in the relative rates of the reactions that give rise to these two compounds can change the final product ratio. Below is a list of molecules that in more than one instance have been found to either speed up a desired reaction or slow down an undesirable one: HMPA, water, N, N-dimethylformamide, N, N, N’, N’-tetramethylurea, urea, dimethylsulfoxide, pyridine oxide, 2-pyridone, N-methyl-2- pyridone, polyethylene glycols, DDQ,  antioxidants, dimethylformamide acetals, 4-dimethylaminopyridine, DBN, DBU,  beta-glycine, EDTA, transition metal salts,  phase transfer catalysts.

For any particular reaction you are trying to improve, most of these additives will have no impact at all. When one of them does exert its effect it can be expected to be effective at low concentrations. For that reason, they can be tested at low concentrations and for this same reason, small groupings of them can likely be tested together.  Any small packet of these additives should be chosen so they do not belong to the same general types. If a group of additives shows some activity in changing the important products ratio only then do experiments need to be done to deconvolute the group to discover which cause the improvement.

For example, I have divided the potential catalysts into five groups, any group of which could be included at a low level in a trial reaction to look for any significant change in the product: byproduct ratio.

A: EDTA, N-methyl-s-pyridone, polyethylene glycol, pyridine oxide

B: DBU, N, N-dimethylformamide, 2-pyridone, transition metal salt

C: Antioxidant, 4-dimethylamino pyridine, dimethylformamide dimethyl acetal, urea

D: beta glycine, N, N, N', N'-tetramethylurea, tetrabutylammonium chloride, water

E: DDQ, DMSO, DBN, HMPA

Probably a better idea is for the experimentalist should consider possible modes of catalysis and choose a group made up of the most likely candidates.

When trying to improve a procedure, improving a method using an additive needs to be considered as much as completely replacing that reaction. 

Acid Traps & Acid Acceptors for Use in Organic Reactions to Remove any strongly Acidic Coproduct

 

If a chemical reaction produces as a co-product a small strongly acidic molecule, the equilibrium of that reaction can be disturbed towards a more complete reaction by trapping that coproduct. 

Acid acceptors and acid traps typically neutralize or otherwise deactivate acid that is already present or that has been created as a coproduct in a reaction. By ‘tying up’ any acid co-product any reversible reaction that involves it is driven to completion and prevented from reversing following Le Chatelier’s Principle.

The following agents have the potential to do this:

Pyridine

2,6-Lutidine

2,6-Di-t-butylpyridine

Poly4-vinylpyridines

Urea

Tetramethylurea

Acetamide forms HBr complex the is insoluble in many organic solvents

Magnesium oxide

Sodium acetate

Calcium carbonate

Mercuric oxide

Ammonium  acetate

Ammonium formate

1,2-epoxy-3-phenoxypropane

Ethylene oxide

Propylene oxide

Epichlorohydrin

Ethylenediamine

Pentamethylpiperidine

1,8-Bis-dimethylamino-naphthalene (proton sponge)

Hexamethyldisiloxane

Trimethylsilyltrifluoroacetamide

Trimethylsilylurethane

4A Molecular sieves

Pyridine ion-exchange resin

Amino Acid Hydrochlorides to Give Zwitterions

N-alkyl-2 halopyridines, formamide acetals, 4A Molecular sieves

Silver oxide

Lead Hydroxide from Acetate: neutralizes acid and gives insoluble salts

Typically acid traps are not bases themselves although all that is essential is that they be substantially less nucleophilic than the moiety that is required to be the nucleophile in the planned transformation. If the nucleophile which is the hoped-for participant in the chemical reaction is readily available and inexpensive the same species can serve in appropriate quantity both as nucleophile and acid trapping agent.

Sometimes the reaction that forms the acid co-product does not involve any nucleophile. Free-radical bromination comes to mind. In this instance, the acid trap must be not nucleophilic at all because it is going to be the strongest Lewis base in the mixture. A substance that forms an insoluble adduct, or which itself is insoluble, or which constitutes a separate phase, or which irreversibly reacts with the acid is desirable. Proton sponge, pentamethyl piperidine, urea, N,N,N',N’-tetramethylurea, and 2,6-di-t-butylpyridine are not nucleophilic.  Magnesium oxide and calcium carbonate are insoluble in organic solvents. Ammonium acetate and ammonium formate are inorganic neutralizing salts that are volatile and so can be pumped away under a vacuum. The four epoxide agents react with hydrogen halides by opening to halohydrins. 

Hexamethyldisiloxane, trimethylsilyltrifluoroacetamide, and trimethylsilylurethane convert free hydrogen halide into trimethylsilylhalide. For this reason, they must be used in sufficient quantity to persilylate any other reactive functional groups in the molecules. Molecular sieves and basic ion exchange resins work by taking free acid into a different phase.

I have used a slightly different idea to remove excess HBr in a process step developed during my industrial career. I added t-butanol to my reaction. It rapidly reacted with HBr as it was formed giving water and non-acidic t-butylbromide.

Glycerol: A Possibly General Method for Changing Solvents in Process Chemistry Reactors

In the laboratory completely changing the solvent that has dissolved within it any mixture of solutes to a different solvent (so long as the solute mixture is stable under the conditions of concentration ) is to completely remove the first solvent on a rotary evaporator, add the second solvent and stir to redissolve any solutes that have precipitated.

This is not even conceivable in industrial reactors because, in almost every case,  they have an appreciable non-stirrable volume. That is to say, when the solvent has not yet been completely removed further evaporation becomes impossible because the impeller blade no longer contacts the liquid. A further problem is that precipitating solutes would accumulate on the walls of the reactor making heat transfer impossible without solute degradation.

 

A possible rather general procedural solution for this difficulty may exist. 

1.    The first reaction solvent is completely distilled (chased) out of the reactor and away from the mixture of less volatile solutes by adding the minimal stirrable volume of glycerol. Glycerol (bp. 290 ℃ with decomposition).

2.    Since neither acetone nor methylene chloride are miscible with glycerol and they do not form any binary azeotrope

 (i) A mixture of some useful proportion of acetone and methylene chloride is added to the glycerol solution or slurry to deliver an upper phase that upon stirring extracts the required non-volatile substrates and the lower glycerol phase is cut away.

or (ii) the desired second solvent, if it is less dense and also immiscible with glycerin, is used to extract the non-volatile substrate, and the lower glycerol phase is cut away.

Then

If (i) above is used, distill the methylene chloride away from the acetone leaving an acetone solution.

If (ii ) above is used, then wash the desired second solvent solution with water or brine to remove traces of glycerin.

Then

If (i)  above has been followed, add the new second solvent and distill away the lower boiling acetone (almost all solvents have higher boiling points than acetone). If this second solvent is to be both protic and miscible with water, residual glycerol needs to have been removed at the stage of the acetone-methylene chloride extract by passing it through a plug of silica or alumina.

Note: For this to work, the mixture of less volatile solutes (usually the entire contents of a reaction mixture ) must be stable in glycerol under the conditions required to evaporate the first solvent and at least the desired components extractable back out of the glcerol.