High-throughput synthetic route screening

[smiley_boy]

I wanted to bring up this subject, because I think it has a great potential for radically advancing phenethylamine and tryptamine production.

Right now, combinatoral chemistry is used by big drug companies to screen through libraries of thousands to millions of compounds to find drugs useful for treating various ailments. The general procedure is they identify what chemical qualities their after, build up a database of compounds that might be interesting, then screen through them to find what they're looking for, narrowing their search down to the desired one or two compounds. In the process, they build "libraries" of compounds by synthesizing thousands of compounds at once (using robots, etc.)

Well, we have a different problem. We have one or two desired compounds, but we're all looking for the most desirable synthetic routeleading towards these.

How might we use this approach and, in a way, help pioneer a cutting edge branch of modern chemistry? What sort of quantitative analytical methods could we come up with to measure the kinetics, etc., of say, variations of the Wacker oxidation, or reductive amination? For that matter, only a tiny amount of literature exists on applied novel routes to 2C-B -- this could really be something.

Does anybody out there have some thoughts on this subject, or even perhaps some literary recomendations?

twitchin' the night away...

[oxycodfish]

I love it!  I've actually been thinking about that for a little while too -- why let the drug companies have all the fun?

It seems to me the best method will be a survey of literature relating to the desired synthesis followed by the design of an exhaustive experiment which will test many varieties of the documented methods.

Let's say we want to test some specific substituent addition.  This "method" would suggest researching many different mechanisms by which this has been achieved, trying to find a pattern among all the methods (in other words, figuring out what variable(s) may be affecting the success/desirability of the reaction), and then continuing to change that variable in a series of experiments.  In this case, we'd run a battery of tests with, let's say, different reagents, at different temperatures, in different solvents, and with different catalysts.  Yes, that's a lot to analyze.  But odds are, each and every one of those combinations hasn't been tried before.  (Obviously, combinations which HAVE been verified as non-working or non-desirable need not be tested.)  Rate testing during the reactions followed by chemical and IR/NMR analyses of products would yield a hell of a lot of data, a lot of reactions which can be dismissed immediately as useless (wrong product), too slow, terrible yield, etc.  Perhaps one or two which merit further inquiry; just one new method in hundreds of experiments would make all the work worth it, I would think.

That's just my suggestion, of course.  It would be great to hear feedback on this or any other ideas.  Why don't we see if we can pick out a problem and do some work on it?

Nobody reads my posts  

[smiley_boy]

Yes, yes! This is what I was hoping for.

Well, let's start with a simple example: the Wacker oxidation. We can vary, among other things, reaction temperature, concentration of reagents, choice of reagents, choice of solvents, and choice of catalyst. What we need to look into are high-throughput methods of analysis in a clandestine setting -- how will one determine reaction completion, and quantitative analysis of final product, of, say, 50 experimental samples running simultaneously? This is the approach we should take.

[Lilienthal]

There is a method which is superior to the independent optimization of different reaction variables.

Choose conditions which make a good fill of the multidimensional space of conditions.
Then add the yields from the experiments. Now you may calculate 'contour lines' or 'iso-yield-lines' and identify the peaks ( = optimized reaction conditions).

There are statistical programs which do the calculation and aid in finding the starting conditions for such experiments (I only read about this long ago...).

[Osmium]

Easy for the wacker. Dilute a drop of the reaction mixture with some solvent and inject into a GC. If the GC analysis can be done isothermally you can inject a sample every 5 ot ten minutes. Inorganic catalyst amount is low so it will all deposit at the beginning of the column which can be cut off when it gets plugged.

I've seen a nice IR spectrometer recently. It is so small that it fits into a reaction flask, and you can follow the course of the reaction inside. Of course it's only usable with certain reactands and solvent systems, but nevertheless a nice toy.