It’s tough to be a graduate student. If you are a Professor, you can sit in your ivory tower and think about all manner of nutty ideas and, as long as none of them violate any laws of thermodynamics, they will be eventually reduced to practice (and improved!) by our capable graduate students and postdocs. But the devil is in the detail and all those brave souls are left figuring out how to lower the kinetic barriers of reactions we contemplate. There are reactions (unfortunately, a lot), in which there are just too many energetically similar pathways, which is why we get in trouble with by-products… Apart from this insignificant detail (I am being sarcastic), chemistry is deceptively simple: any idea about an isolable endpoint of synthesis that is not uphill in energy, is worth the risk. Needless to say, you can design special conditions and isolate uphill intermediates (e.g. carbocations), but this would amount to imposing a kinetic barrier of some sort. Now, are there ridiculous (but seemingly plausible) ideas out there that can throw us for a loop unless we sit down and think about them for a second? Here’s one of the problems I like to discuss with my colleague, Professor Jik Chin. Consider the following generalized process:
Imagine that you want to develop a catalyst that would run this reaction. Can such catalyst exist? No, it can’t. The way this reaction is written is sheer nonsense. For this conversion to have a chance to work in the forward direction, the Gibbs free energy change must be less than zero. In the example above we clearly have no entropy change and enthalpy does not change either. In addition to the violation of the Second Law of thermodynamics, there is a problem with the principle of microscopic reversibility here as any catalyst that works in the forward direction should be capable of catalyzing the opposite process. Of course, stoichiometric reactions can be designed and there are many solutions for this “R into S” type of problem. Enzymes can do this too (and catalytically!), but those reactions are coupled processes, which means that there is something else that goes on with either your product or your starting material. Hence, the energy of the product is not the same as the energy of the starting material. You can break microscopic reversibility with photochemistry, but if you are interested in thermal activation, any catalyst that you think might promote the aforementioned process, will necessarily have to violate microscopic reversibility and the Second Law. Back to my starting point: unless we propose thermodynamically ludicrous ideas, being a Professor is the best job out there. The way George Olah would say, “Hey – I am doing my hobby and the University even pays me for it!”
amphoteros 
This is also the reason why it is fairly rare for racemic compounds to spontaneously crystallize as a conglomerate of crystals of single enantiomers, (only 1 to 3% of all racemic compounds do this.), usually racemic R+S 1:1 co-crystal forms.
It is important to explain to students that any reaction candidate to be turned into its asymmetric version must be completely irreversible, otherwise the initial ee erodes over time due to the equilibration. Racemate is always the thermodynamically favored product, by a factor -0.693RT. So, a normal version of aldol condensation reaction is not suitable for chiral catalysis but Mukayama aldol is
where do you get the 1 to 3% figure from? Do you have a reference?
I wonder myself. Hope milkshake can clarify…
If I remember correctly, it was in the Jacques-Collet monograph (“Enantiomers Racemates and Resolutions”):
http://www.amazon.com/Enantiomers-Racemates-Resolutions-Jean-Jacques/dp/0471080586
I remember that discussion in the book went something like this:
For racemic conglomerate formation to be thermodynamically preferred over a 1:1 racemic co-crystal, the enthalpy advantage of a conglomerate crystallization (over a 1:1 co-crystal) needs to be higher than the entropy cost of a spontaneous resolution – a serious handicap. Salts and other highly polar compounds have somewhat higher frequency of conglomerate formation when compared to less polar compounds (because the stronger interactions in crystal, the higher is a chance that the enthalpy difference of the two crystal forms will be enough in the direction of the conglomerate) but even then, the conglomerate frequency is only about 3%
I could be wrong about this number; it has been 25 years since I was reading the monograph for my class.
(The blog won’t let me reply after you last comment)
On pg 81:
Of 1308 chiral crystalline compounds from the Beilsein Handbook, 126 compounds had higher melting points (at least 20 ºC) for pure enantiomer vs racemate. They prepared 32 of these 126, and 21 were actually conglomerates. From this they estimate the frequency of organic (non-salt) conglomerates at 5-10% and admit it is not very precise.
They estimate that in salts, conglomerate formation is 2 to 3 times greater than in covalent compounds.
Thanks, this is good!
Thank you – my numbers were way off!
This is an excellent point, I agree!
I do like your blog very much – always stimulating!
Thanks!
Great topic Andrei!
nice and relevant review: http://onlinelibrary.wiley.com/doi/10.1002/anie.200804566/abstract
You can “beat” the principle of microscopic reversibility by some well established tricks:
recent one:
http://pubs.acs.org/doi/full/10.1021/ja4082827
and maybe the most famous one was by D. J. Cram and his “amino acid resolving machine”
Thanks for bringing these paper up to the forum – they are great references, I agree. Still, though, the principle of microscopic reversibility stands solid. But I agree that there are some clever workarounds. See – in Dean’s case the trick is phase separation… The classic U-tube experiments by Cram also rely on a “shift”. Very clever stuff, though.
Pingback: Blogroll: The process : The Sceptical Chymist