“Development of safe and scalable continuous-flow methods for palladium-catalyzed aerobic oxidation reactions” Ye, X.; Johnson, M. D.; Diao, T.; Yates, M. S.; Stahl, S. S. Green Chemistry, 2010, 12, 1180-1186. DOI: 10.1039/c0gc00106f
We’ve had a pair of posts recently about using oxygen as an terminal oxidant in cross-coupling and biomass degradation, and as a green oxidant, it’s pretty hard to beat. So I was a little surprised to learn that of the many cool aerobic synthetic methods that have been developed in the last decade, very few are used in industry. The big drawback, especially on large scale, is safety – oxygen is usually the limiting reagent in the combustion reaction, and things can get pretty crazy when you have an oxygen-enriched atmosphere (and much crazier with liquid oxygen – check out this awesome video, and this one that Marty had in his last post). So while stirring 100 mL of toluene under a balloon of pure oxygen might be fine, doing the same thing with 100 L is problematic.
Safety aside, these reactions suffer because proper gas-liquid mixing is more difficult to achieve as you scale up. All of this prompted a collaboration between Eli Lilly and Shannon Stahl‘s lab to develop a scalable continuous-flow method for aerobic alcohol oxidation, which avoids these problems.
[An aside: Oxygen is used industrially for the synthesis of some chemicals on pretty huge scales – this paper lists 15 chemicals produced at million ton/year levels by aerobic oxidation. On that scale, it makes sense to specifically design reactors that work safely and efficiently for each process. But for smaller scale chemicals, like pharmaceuticals, it’s not worth it to design a new reactor for each oxidation, and most processes are conducted in stirred-tank reactors that aren’t suited for handling gaseous reagents.]
The authors decided to look at a fairly simple reaction, the oxidation of alcohols to yield the corresponding carbonyl compound. This transformation is typically carried out using chromium reagents, which are extremely toxic (but still taught to undergraduates for reasons I don’t understand), or reagents based on hypervalent iodine or activated DMSO. The aerobic version, first reported by the Sigman lab, is catalyzed by palladium acetate and requires an atmosphere of pure oxygen.
The catalysis proceeds through a Pd(II)-Pd(0) cycle, with oxygen reacting directly with the Pd(0) species to convert it back to Pd(II). The main issue with this reaction is that the Pd(0) will precipitate out as metallic Pd if it isn’t oxidized quickly enough, a problem that occurs when oxygen concentration is low and at elevated temperature.
In the original oxidation paper, the reactions were set up by combining reagents and solvent in a flask under a balloon of oxygen, and then stirring until the reactants were converted to product. In a continuous flow process, reagents and solvent are injected into a metal tube, and the reaction mixture is pumped through the tube until it comes out the other end. At this point, hopefully, the reactants have been converted to product – if not, the residence time is increased. The small reaction volume and movement of the reaction mixture through the tube produce very efficient mixing, which can reduce reaction time. This also makes it easier to control the reaction temperature, which is challenging in large batch processes. As far as I can tell, it works something like an HPLC, with pumps that can control the quantities and flow rates of different reactant solutions. The main difference, I think, is that the pressures involved are lower, and longer coiled tubes are used. Here’s a picture of what I believe to be a typical small-scale setup (taken from this paper, but please let me know if you’ve got a better picture that I can use!).
Crazy, right? But it works pretty well – only slight modifications to the reaction conditions were needed for the reaction to work well in the flow reactor. Catalyst decomposition isn’t a problem in the flow reactor because of the improved gas-liquid mixing, so they’re able to increase the temperature, reducing the reaction time from 18 hours to a residence time of 45 minutes. They could run the reaction under 30 psi (2 atm) of pure oxygen, or they could use diluted oxygen in nitrogen, as long as the partial pressure of oxygen was 30 psi – this allowed them to run their reaction using 8% oxygen, a level of dilution at which there is no great danger of explosions.
The process was scalable too – switching from a 400 mL flow reactor to a 7 L flow reactor required very little modification of their reaction conditions, and allowed them to run the oxidation on a kilogram scale in quantitative yields. The paper mentions that reactions have been successfully performed on the 100 kg scale in the 7 L reactor, a scale that would require over 1000 liters of solvent under the original Sigman conditions. So the flow reactor has an additional advantage in requiring less space to conduct the reaction on a large scale. All in all, this is a really cool example of how engineering can enable the adoption of greener reaction conditions. Apparently there are other ways that continuous flow processing can make a processes greener, although I don’t really understand them yet – maybe it’s something worth blogging about in the future!