How thinking about Separating Acetic acid and Acetic Anhydride can sharpen our Chemical Process Development Skills
kilomentor | 16 December, 2011 10:50
Ganesh Wagh in the Chemical Process Development Group on LinkedIn asked the question, “What is the best way to separate a mixture of 60% acetic anhydride and 40% acetic acid to recover the acetic anhydride?” This solicited quite a few responses. The particular forum is now closed and although Kilomentor did contribute one suggestion to the forum I think the topic deserves some further discussion that doesn’t fit well into the question-comments format.
A plurality of people suggested either distillation of distillation under reduced pressure. These solutions may very well be the best if one has the proper fractional distillation equipment and it is not busy performing more valuable separations, but there is the implication in the question itself that, at least in the questioners hands, there was some difficulty with these solutions, since they are the first thing to come to a process chemist’s mind. Indeed Ganesh later noted that he did encounter problems in his hands.
Several people (Antonio Manka, Anton Frenkel, Marta Krawczyk, Pravinchandra Vankawala) suggested fractional distillation, either with some variation of the distillation condition for atmospheric pressure or under reduced pressure. I am not an expert on distillation. That is perhaps why it is not clear to me why distillation at reduced pressure would be expected to simplify separating these two liquids compared to an atmospheric pressure fractionation. Also, whatever the pressure under which the fractionation is done, these methods appear to require the complete removal of the acetic acid before the acetic anhydride distillation and that must leave some lost still bottoms and column residues that must subtract from the recovery. On top of this, I would imagine that fractional distillation must be slow and labour intensive. One suggestion, variation of the packing in the distillation column,which would change the height equivalent of a theoretical plate, is typically not readily accessible in a multipurpose plant particularly where the object is no more than reagent recycling.
Mark Bratt contributed the idea “Any thoughts as to a non-aqueous separation e.g. cyclohexane or methylcyclohexane with methanol or acetonitrile, with an organic base?” Although I don’t endorse trying a non-aqueous, two liquid extraction that uses methanol, which would be both miscible with and reactive towards acetic anhydride (this is almost certainly just a slip – it is true that methanol is very frequently a partner phase in such applications), the idea is suggestive of a type of strategy that is adapted and taken up by others later in the discussions. Perhaps we could simplify this idea. What would happen if we just added enough triethylamine to neutralize the acetic acid? Wouldn’t the triethylammonium acetate form a separate liquid phase aside from the acetic anhydride?Even if it didn’t, wouldn’t distillation of this mixture amount to an reactive distillation in which the acetic acid would be held back by its reversible interaction with the triethylamine allowing the acetic anhydride to distil over cleanly first? Acetic acid and triethylamine are reported to give a higher boiling azeotrope, with composition 69% acetic acid 31% triethylamine, that distils at 163 C. If triethylamine didn’t do the job what about neutralization with tributylamine?
Kees Hoek made a suggestionto add acetyl chloride that would, if it is worked, convert the acetic acid back to acetic anhydride thereby, in principle, multiplying the acetic anhydride recovery. My rough bond energy calculation for this reaction between acetyl chloride and acetic acid gives an enthalpy difference of close to zero so if the kinetics were practical it most likely would produce an equilibration of starting material and products. Driving out the hydrogen chloride would be expected to drive the equilibrium towards acetic anhydride according to LeChatelier. The usefulness might boil down ( pun intended) to how fast one can scrub hydrogen chloride. It might be attractive if one also had in the plant an alkaline waste that needed neutralizing.
Pedro Guivisdalsky, among other ideas, suggested adding toluene to form the binary lower boiling azeotrope with acetic acid that distils at 105.4 C. This is a variant on the extractive distillation idea whereby the acetic acid is made relatively more volatile. Waste toluene could be used and the toluene/acetic acid washed to recover the toluene.
John Knight, Scientific Director at Scientific Update, suggests neutralizing the acetic acid using the completely non-volatile basic ion-exchange resin. As he comments, this risks the exotherm that accompanies every mode of purification that features a neutralization. It would also involve bashing around a lot of insoluble resin that may be ground down and give some tedious still bottoms. Nevertheless, the core idea is that insoluble nonvolatile ionic exchange resins should come to mind when neutralization is considered in separations generally. Also, Dr. Knight again relates back to the idea of using some agent to convert the acetic acid to acetic anhydride. “What about phosphorus pentoxide”, I am musing to myself? The coproduct will be non-volatile and can be rinsed out of the still-pot after careful hydrolysis.
Many contributors to the discussion recommend looking at how acetic anhydride is made commercially as a pointer to an economic method. This is of course excellent advice since the cheapest starting material for acetic anhydride must be acetic acid. The methodology that is used for commercial product may of course be out of the question for recovering usable acetic anhydride in a multipurpose plant where the cost reduction for an accompanying process may be the actual motivation for the question.
Ilya Avrutov seems to have garnered the most endorsements for his suggestion that the acetic acid in the mixture could be made non-volatile by treatment with sodium carbonate, preferably anhydrous sodium carbonate. Others would include a drying agent. All understand the problem that the carbonate leads to water that must be taken into account. Ilya’s concept is another variant of a non-aqueous neutralization like that of Mark Bratt, above; it is simplified by not requiring further organic solvent additions but complicated because the coproduct of his neutralization can react with acetic anhydride to regenerate acetic acid.
This in turned stirs up another pertinent question. Are there cheap agents that can neutralize acetic acid without producing something that would react with acetic anhydride? Two chemicals that come to mind are calcium hydride and calcium carbide. These substances should come to mind whenever one wants to generate calcium salts of carboxylic acids for isolation or derivatization. In this instance the coproduct with calcium hydride would be hydrogen gas which would be innocuous but needs to be treated carefully. Using calcium carbide the byproduct would be acetylene gas that needs to be recognized and handled safely. In each case exactly how these solids are contacted with the acetic acid/acetic anhydride mixture would need to be worked out carefully. I suggest adding either as a slurry in more acetic anhydride.
I think I can still take credit for the most off-the-wall suggestion in the discussion. It was based on a simple question. What is the relation between the water hydrolysis of acetic anhydride and pH? Couldn't one stir the mixture of acetic anhydride and acetic acid with an aqueous salt solution buffered at the most unreactive pH and quickly extract the acetic acid into the aqueous phase? Stirring should be gentle to reduce interfacial contact and the temperature should be kept low to minimize reaction, and of course, separation of the layers should be done promptly. The idea here is that the difficulties that often stop us from reaching a simple solution simply may be mistaken. Here we assume that acetic anhydride will react with water and make a water wash unthinkable. But how well does this reaction actual take place and can it be slowed down enough to perform washing? Remember the Hinsberg reaction of a sulfonyl chloride is performed in contact with an aqueous alkaline solution!