If the KiloMentor blog were being written twenty years ago it would be different. If it had been written thirty years ago it would have been different again. The progress of a technical art, such as process development, creates dramatic changes. The step which had been a bottle-neck in the creation of a process becomes less demanding and another aspect of the art becomes the chief challenge to the scientist. Thus, if any of these documents were revised in another dozen years, the relative difficulties of different aspects of the challenge will have again changed and the reasons for the proposals made here may have evaporated and the advice provided may become completely wrong-headed. With this in mind, an author should at the outset state what the status quo is in his field at the time of writing so that future readers can decide for themselves whether the same state of affairs still exists and if not what logical changes should be inferred in the recommendations being offered.
To illustrate this point, forty-three years ago, when I was just beginning my activities as a process chemist, devising the sequence of organic reactions that could lead from commercially available starting materials to the target product was the great challenge. Dependable reactions were limited. HPLC did not exist. IR and UV measurements were just being replaced by NMR methods for following reactions and identifying products. But the more important difference was that on-line database searching did not exist. Most significant of all, electronic substructure searching did not exist. A close analog of the relevant portion of a molecule you wanted to synthesize might be present in the literature, but there was no dependable way to find it.
To-day, although many would disagree about the degree of change, creating a realistic synthesis scheme for substances of moderate complexity is no longer the most challenging step simply because we have replaced our memories and our punch cards with computer memories professionally indexed available on our desktops.
In this earlier period, when creating the process step flow sheet was the dominant challenge, we synthetic chemists got into the habit of measuring excellence in synthesis by measuring the number of sequential steps in the longest arm of the contemplated process or the total number of reaction steps. We also calculated the overall yield although this could only be determined after the proposed sequence was converted to a real process.
Because this challenge strained our ability to approximate we made some rough assumptions in inaccuracies of which we hoped would balance out between competing processes and still allow us to make a valid judgment of which was the most promising. For example, we assumed that the amount of crude impure product which one got out from a reaction step was proportional to the quantity of product, which was present in crude form in the mixture after the reaction was complete. Why do I say we made that assumption? Because we looked at the yield of model reactions for simple substrates with just one or two functional groups and assumed that the recovery of pure product from these would be about the same when we used complicated substrates with multiple functional groups. In the terminology I will use herein, we assumed that the isolation yield- the yield of pure product as a percent of the assay yield (the amount of product in the reaction mixture) is consistent for a reaction type. The corollary of this which we also adopted as a simplifying assumption was that isolation although it might be tedious was routine and could be taken for granted as generally of similar difficult when integrated over all the steps of a process. That is isolations can be ignored so long as the total reaction steps are minimized.
Today, the focus of attention has shifted. Scientists can gather large amounts of relevant data about the likely properties not just of substances that have been reported somewhere in the vast chemical literature, but even predicted properties of unknown molecules. What the electronic databases have not been able to do is help us select the simplest and most rugged purification methods to use with reactions at scale. What we as scientists have not tried to do is ask ourselves the question, “What kinds of functional groups do I want in my intermediates because they will simplify the isolation and purification of that step?”
An axiom of the approach to process development that will be found here is that there is an advantage in ranking the degree of difficulty of isolation/purification of each process step and using it as an additional criterion of selection of the most preferable paper chemistry route along with the traditional criteria: number of steps, number of steps in the longest branch of a convergent process, and the known approximate yields for the reaction type. The result of this I predict will be that preferred processes may on average contain more process steps but the speed with which these steps can be carried out will be much higher, the overall purity will be much higher, and the cost will be much lower because the time spent in the isolation purification is typically much more than the actual reaction time.
No comments:
Post a Comment