A Question from the LinkedIn Organic Process Research & Development Group that Fosters Practicing Creativity

kilomentor | 14 January, 2012 15:24

A question was asked at the LinkedIn Organic Process Research & Development Networking Group http://www.linkedin.com/groupItem?view=&srchtype=discussedNews&gid=1902030&item=88459764&type=member&trk=eml-anet_dig-b_pd-ttl-cn&ut=07TQuSi7kXnl41 “How could one remove a small amount of ethoxytrialkylsilane from a solution of ethanol containing some water at a scale of a large commercial waste stream.

I do not know whether a good answer is possible for this question. What interests me is what instruction the question can provide to students of process chemistry about how one searches for possible constructive leads for research.What is the entire real question?

Larry Fertel who posed the problem says, “I've got a stream of ethanol contaminated with <0.09 wt. % of a low boiling, trialkylethoxysilane. Any ideas on how to remove this (adsorbtion, coordination with a flocculent, etc.)? Distillation is not feasible, as the b.p.s are similar. Note that this is on a commercial scale….There actually is 9% or so water in the stream, water content is not an issue….The problem/issue is that the silane has a b.p. within 2 deg C of the azeotrope, sorry for leaving that piece of info out….Let me say that the ethanol stream is many millions of gallons per month, so this is not a research project. The ethanol is a by-product of a commercial process, so replacing it is not an option. Any treatment must work within the economics of reselling the ethanol into the marketplace (i.e., can only cost so many cents/lb.) . One option we are looking at is to convert to ethyl acetate for reselling. We are considering carbon treatment, one can build large columns in parallel that can be used on a continuous basis and get switched around once the carbon gets saturated in the impurity (assuming it works)…. I work for a contract research company, we were approached by a customer with this issue (good for us ($$), bad for them!!) So we need to be the problem solvers. I worry about the kinetics of adding a small amount of a reagent to a small amount of impurity. Considering that it is a bimolecular reaction, wouldn't the rate be slow? Not sure of the spec needed for the final product, at this point we are at the "screening" level.”

The problem is that there is <0.09% of an unacceptable compound contaminating millions of gallons per month of ethanol/water which cannot be removed by distillation because the undesired trialkethoxysilane distils close to 78.2 C, the ethanol/water azeotrope.
The problem is not a solution of ethanol/water. It is the presence of the trialkylethoxysilane at the level of <0.09%. There must be some low level that is acceptable otherwise the problem is to reach an undetectable level at low cost and that is almost certainly impossible. We must assume that the actual level is around 0.09% and it needs to be substantially reduced to let us say 0.01%

The problem is that the process as designed gives this level. It is not that that the level is around 0.09% because of some deviation in the process. There was no time when the level was acceptable. The unacceptable level is always present in the exiting waste stream from the process.

To solve the problem we might argue that we cannot remove the ethanol/water from the stream. That is we might say we cannot act on the major content of the stream because that would be expensive but apparently that is not the case. Larry Fertel says that distillation is rejected because it does not work not because it is too expensive. This opens the possibility that if we could selectively convert the trialkylethoxysilane to something separable from ethanol/water it would constitute a possible solution. Similarly it would seem that somehow changing the distillation temperature of the silane would be a possible solution and it would be particularly advantageous if it were the impurity that vaporized at lower temperature. Adsorption or adsorbtion onto or into another phase would be advantageous and probably most preferred because the predominant ethanol-water would be unchanged.

The hydrolysis and adsorption ideas have been pretty well explored by Stone, Tyrell, and others in the LinkedIn discussion. Kilomentor would like to look at the distillation and reaction possibilities a bit more.

One way to increase the degree of separation between the points of distillation of the trialkylethoxysilane and the ethanol/water is to add something to the mixture which will enhance the rate of vaporization of the more lipophilic material. This is the idea of extractive distillation. At the same time one does not want to leave anything behind in the ethanol/water that you are trying to decontaminate. Suppose we were to add an essentially immiscible perfluorocompound (expensive) or a much more volatile pentane (cheap). Would the vaporizing added compound sweep with it the undesired silane? I am reminded that there is a patent that claims the removal of traces of DMSO by co-vapor entrainment with ??.
It has already been suggested that a hydrophobic styrene-divinylbenzene cross linked resin might work. What about just using beads of solid paraffin? These would be cheap, sturdy and insoluble. They could be created by dropping molten paraffin into the stirred bulf solvent. They would float on the surface of the ethanol/water and could be filtered off and melted to drive off the impurity.

Remember guys, during the idea generation stage of problem solving criticism is supposed to be put on hold!

Now what can we get by playing with the idea of reaction of the impurity? What distinguishes the undesired impurity from the bulk? Molecular weight. The presence of silicon. The presence of a silicon oxygen bond. The presence of carbon-silicon bonds. The presence of silicon in the impurity seems to jump out at me. What is the strongest bond that silicon can form because that is what I want to do if I am going to drive this product into some other substance? The strongest silicon bond is one to fluorine. So perhaps the ion exchange resins that are being contemplated should have fluoride ion loaded on them. The conversion R₃SiOEt + F^-+ H₂O going to R₃SiF + H-OEt + OH^- would have an enthalpy advantage of about 27 kcal/mole. Hydrolysis in contrast has no enthalpy advantage, I think.

Anyway, problem solving is a combination of a good statement of the problem, a non-critical atmosphere for generating ideas, and most important well loaded retrievable associative memory banks. Note that you cannot come up with good ideas based on stuff you can’t recall but need to look up. Looking stuff up can be used to refine ideas but it cannot help to create them. You need to know things well enough that they can be recalled by associative memory when presented with the clues offered by the problem itself.