Wednesday, January 18, 2012

Criticisms of "Atom Economy"

When I wrote about "atom economy" yesterday, I tried looking around for some serious criticism of the approach and didn't find much, so here are my thoughts.

Atom economy is defined by Barry Trost of Standford (subscription/pay-per-view) as the molecular weight of the product divided by the sum of the molecular weights of the reactants.

Chemists have in the past relied on yield percentages to measure the efficiency of a reaction. If there is 10 millimoles of A at the start of the reaction and it is transformed into 9 millimoles of B at the end, the efficiency is 90%. These yields are important, as many reactions are part of a multi-step synthesis. If a 5-step reaction has 90% yields for each step, the overall yield is 0.95 = 59%.

With that 90% yield however, there may be a mountain of chemicals, solvents and catalysts left in the wake of the reaction and that shouldn't be ignored. In one of my earliest jobs working with coating solvent-dispersed adhesives, we were all very much aware that for every 3 gallons of adhesive brought into the plant, 2 gallons of solvent (plus a lot of heat) went out the roof of the building to the thermal oxidizer, so certainly a "big picture" approach to chemical reactions is needed. It's just that I think atom economy is a lousy option.

  • The impact of atom economy is greatest for large volume, industrial situations, and not small academic labs. However, much of the research published on atom economy concerns academic research rather than industrial preparations. While in some cases, the difference between the two is mostly a matter of scale, in other cases, industrial processes are whole different than research lab processes. Worrying about the atom economy of a non-industrial reaction is a waste of time.
  • By focusing solely on the nature of the reaction and not the actual chemicals involved, atom economy overlooks process concerns such as ease of separation of the product, the hazardousness of the reaction and the byproducts, etc.
    • It inherently favor reactions without any byproducts (additions and rearrangements) and disfavors all others (substitutions, condensations and eliminations).
    • It also favors the upper rows of the periodic table. Lower rows, while often being more reactive, have greater atomic masses and are therefore disfavored by the calculation. Yet many preparations of fluorinated compounds are far more dangerous and hazardous than the chlorinated equivalents. Is this not a concern?
  • It is often unclear exactly what "the reaction" is. Taking the example again of (poly-)amide formation between an acyl chloride fluoride (lower MW in the reactant = GOOD!) and an amine. I'm not aware of any naturally-occurring acyl fluorides, so that material was prepared at some earlier point. Shouldn't that reaction be included in the overall calculation as well? And what if the amine also required preparatory reaction(s), shouldn't that be included in the atom efficiency calculation as well? How far back should this all be pushed? Is cradle-to-grave the proper approach? Or can we game this to a better starting point? (I find it rather ironic that the modern organic chemical industry itself is a "byproduct" of the petroleum industry. If the refiners had practiced better atom economy, modern organic chemistry would be much in arrears from where it is today.)



Just like yield percentages, atom economy looks at only one aspect of a reaction under the guise of being "green". A truly green approach however, requires some thought and critical analysis, not just a number that can be reported on a scorecard.

1 comment:

milkshake said...

Amide couplings are one of the least problematic bond-forming reactions in general but anyone who worked on peptidomimetics and or in combichem can confirm that combination of hindered substrates and especially acylation of amines with electron-withdrawing substituents is a problem. I remember using a combination of acyl chloride and several equivalents of silver cyanide in DMEU in one particularly nasty acylations, this being the only one method that worked. Yet another problem is with unstable substrates that crap up during activation/coupling.

With regards to atom economy, I find it disingenuous when people obsess about the molecular weight of a reagent while at the same time running large-column chromatographic purification of their products, a method that consume gallons of organic solvents per gram of purified product. Also I remember once using triphenyl phosphine for reduction of arylsulfonyl chloride to arylthiol. This requires about 3.5-4 equivs of PPh3 and with molecular weight of 250+ a reduction done on a moderate scale meant that I loaded a kilo of PPh3 into the flask and then had to remove a kilo of PPh3O from the reaction mix, but it was not a major problem once I found a system where the phosphine crap was soluble and the thiol product was not.