[Mitch]

As a scientist and fellow chemist I never understood this behavior. Wouldn't it be best to report the average yield for the reaction and the associated standard deviation? I understand this wouldn't be prudent for a long synthesis that has only been repeated once, but most reactions are repeated many times over.

[tmartin]

To me, this topic opens up a whole can of worms.  I agree that reporting the average yields is often times a much better way to report your reactions, and really the way it should be done.  The 89% outlier when your averages are normally 50-70% just doesn't seem to be the right way to go about reporting things.  Even in synthesis, most steps are done a few times, so it would be best to report the range.  (I'll give that the first time you run a reaction, chances are your yield will be lower and it should improve)

From my perspective, I can get a feel for how you always want more material (to push forward to those precious intermediates at the end of your long synthesis), so you want higher yields, so perhaps in some weird way that comes into our reporting, and by saying "I got a 90% yield" even if it was only once, makes people feel better as a chemist?  Of course, if you're doing a new methodology, you want to tout high yields so people will use it, and in a synthesis you want high yields so people say it is a good synthesis.  

There have been plenty of times when I have seen a very impressive step, that only proceeds in say 50% yield and I think "oh well that's not as good as it could be", but then when I think about "Oh hey they just 4 stereocenters with high selectivity!  " that 50% yield gets a lot better.  I think it's in part human nature to think that 100% is the best, so you'd want to get as close to that as you can, even if it is at time a little disingenuous (which if you think about it maybe is a little sad that things have come to that).

[stewie griffin]

I completely agree with you. Yield inflation is annoying, dishonest, and unethical.
Another trick, along the lines of only reporting the best yield, comes with oxidations. Sometimes I'll see a Dess-Martin oxidation work in high yield, but it appears really early on in the synthesis. Most folks realize that Dess-Martin reagent is expensive, and I am willing to bet that when the synthesis is scaled up that the oxidation occurs under Swern conditions. Indeed, if you check out the experimental you'll often see that the reaction before the oxidation is on several grams, then the oxidation itself is carried out with Dess-Martin on maybe 100-200 mg, and then the next reaction is again occurring on gram scale. Seems fishy (my bet is that the Swern just happens to give a lower yield, so they report the more appealing reaction yield even though it's not how the synthesis was practically carried out).
The problem with yield inflation is that it only takes one group to ruin it for everyone. If my group decided we were only going to report averages then how can we possible compete for funding when other groups can "easily" get 90%+ in every reaction they touch? I must say I like the fact that some of Baran's syntheses have steps with yields in the 40's and even in the 20's (if I remember it correctly). That's great b/c it's honest and although the yield sucked the community still learns a lot from a elegant synthesis. However, how many people will get their papers published in JACS and ACIE with yields below 50% if your name isn't Baran, Corey, Nicalaou, etc..
That's my two cents for now. Gotta go grade...

[Dan]

I understand your point, but I imagine the reason is that the bad yields are usually assumed to be associated with user error or unoptimised procedure - not dry enough, unstable and was sitting around for too long in workup, spilled a column fraction, bumped on the evaporator etc. etc. Facors such as these are then assumed to be experimental anomalies and do not contribute to the mean. As you can never do a reaction exactly the same way twice one can argue that every yield obtained that was lower than the best one is due to a fluctuation in conditions or slight variation in procedure that does not represent the optimum... I suppose an argument could be: Where do you draw the line between an inconsistent chemist and an inconsistent reaction? I don't know...

I suppose the idea is something like "if done rigorously this is a yield you can expect". I agree that an indication of how sensitive a reaction is would be useful and is often not really mentioned.

I report one of my better yields on a reasonable scale - a trade off between scale and yield. Given the choice between a 90% yield on 50 mg scale and 80% on 500 mg+ I would take the latter every time. The Swern and Dess-Martin example posted by Stewie is an example I like and a very good point. In our lab we try to stay away from that, quoting yields for sequential stages on similar scale. If there is a large disparity between different scales it should be stated - for example, in a recent paper for a couple of steps we give two yields: one for 5-10 g scale with column isolation and another for 100-200 g scale with recryst. (which obviously is substantially lower).

[da692]

Heathcock and the old guys used to report yields over a range of attempts. I think it is the proper way to do it. I don't know when exactly the transition occurred but it is easy to assume why it occurred. I have a really hard time accepting any yield for more than face value, especially this 98% crap (after chromatography) you repeatedly see in the literature. Hudlicky addresses this issue in his 1996 chem rev and his book 'the way of synthesis' - pretty good read

[stewie griffin]

da692:
I've read Hudlicky's views on this topic in "The Way of Sythesis". It was awesome to see him call b.s. on many of the reported yields after chromatography. Definitely a good read, and it should be required reading for all organic students. Perhaps it should be handed out in teaching labs so that students understand that it's ok to get a 63% yield.
To be fair, as Dan points out, you do often get better and better at the reaction as you repeat it and go through the synthesis, so naturally the yields get better. But there should still be some restraint and honesty on our parts to give a typical yield rather than just the best outlier.

[dunno260]

One of the sad facts of being an organic chemist is that you have to learn to apply the ole Corey 1,2 shift to suspicious yields.

[MissPhosgene]

Before I knew that yields weren't averaged, I thought I was an utter failure with the opposite of the King Midas touch.