« Testosterone, Carbon Isotopes, and Floyd Landis |
Main
| The Last Word on Eerie Glowing Labs »
August 2, 2006
A Law of the Lab: Yields and Variations
Posted by Derek
The "Law of the Lab" I alluded to the other day is:
Yields go down faster and more unexpectedly than they go up.
My synthetic organic readers will all know what I'm talking about here. We've all had the experience of running a reaction that we've done many times before, only to find it suddenly giving half the yield that it usually does. One of the most important jobs of a process chemist is to iron things like this out, making sure that they don't happen by either tracking down the variables responsible, or ditching the reaction entirely for something more reliable.
But med-chem types like me don't always have enough time to spend on that sort of thing, so we have a lot of reactions that are in a sort of unstable equilibrium with respect to reproducibility. As long as the different factors involved - purity of the starting material, rate of addition of reagents, efficiency of heating, cooling, and stirring, etc. - are within their (sometimes narrow) green zones, things are OK. But let one or more of them wander off, and the fuses start to blow. All reactions will go to pieces on you if you push such variables too much off their mark, but the difference is that a robust one will stand up to all the variations that you'd usually encounter. A wonky reaction is just one sensitive to something that can be over the line under normal conditions.
And there sure are a lot of them. And the different chemistry that starts happening when things cut loose has a far greater chance of messing things up than it has of improving them. Most organic chemistry reactions are very artificial systems - we're using energetic reagents and conditions to make molecules go down particular paths that they wouldn't do to any useful degree by themselves. There are so many other things they can find to do otherwise, and they'll explore those pathways if they get the chance.
So while it's not completely unknown for a random variation to improve a reaction, it sure is rare. Most of them lead to yet another synthesis of the sticky brown gunk which seems to be a universal thermodynamic sink of organic chemistry. You're threading your way through a swamp of that stuff when you do synthesis, and liable to sink down into it at any moment.
Comments (16)
+ TrackBacks (0) | Category: Lowe's Laws of the Lab
POST A COMMENT
- RELATED ENTRIES
- › Dealing With Hedgehog Screening Results
- › Animal Rights, You Say?
- › Blogroll Update
- › Pharma's Return on Investment: Yikes
- › How A Real Drug Industry Project Meeting Goes
- › Ghostwriting
- › Just Give It to NIH
- › How Not To Do It: The Secret Patent Decoder Ring
1. GC on August 2, 2006 10:09 PM writes...
My favorites are the occasional synthesis papers where they report that once they tried to optimize a reaction (and REALLY rigorously exclude water), it goes to hell. They then found out adding 500ppm water to the reaction mixture helps out after all.
Permalink to Comment2. Process Wannabe on August 2, 2006 10:17 PM writes...
I've got a gripe with quite a few total synthesis papers I've read recently. The text quotes a lot of good yields, but the actual experimental section tells a different tale. The first few steps will be on the 50-100 mg scale, with very good yields. All of a sudden, 50 mg jumps up to grams! I find this dishonest. Unless you actually sit there and run 100 of the 50 mg scale reactions to get to your 5 grams, just list the yield of the gram scale reaction, it's what you used to get your material for Pete's sake! This goes along with those finicky variables for a reaction, scale-up can be a wonk.
Permalink to Comment3. Black Knight on August 2, 2006 10:23 PM writes...
Heh. From a biological perspective,
"So while it's not completely unknown for a random mutation to benefit a species, it sure is rare. Most of them lead to yet another synthesis of the sticky brown gunk which seems to be a universal thermodynamic sink"
and you have a pretty good description of natural selection in action.
Permalink to Comment4. Pete on August 3, 2006 12:34 AM writes...
Process Wannabe: I'm willing to bet in a lot of those cases that the small-scale reactions were run just for the analytical data, i.e. with rigorous purification to get nice clean spectra and pass elemental analysis. If the authors are pushing through grams of material at the beginning of the synthesis, it's likely that at some point the material is just "clean enough", so there the yield would be misleading.
Permalink to Comment5. Tot. Syn. on August 3, 2006 4:10 AM writes...
In my experience, the yeild always goes to hell just when you think you've sorted the reaction out. For example, you did the reaction on a 50mg scale, then a 250mg scale, and all was well. Then lump a couple of grams at it, and boom, there goes the yeild. The bugger always catches you out.
Permalink to CommentI've actually just had the trouble of the reverse scenario. One of the first steps in my synthesis is rubbish, and has remained rubbish through most of my runs, *except one*. Yep, one time it went in >85%, and I have no idea what I did differently. The only thing I can suggest is that the solvent I used came from a different bottle - always new, but this time Fluka rather than Aldrich... argh!!!!
6. syn on August 3, 2006 9:25 AM writes...
If you guys were working for my PhD advisor, he'd "kill" you for that kind of unproducibility. I was made to feel like I was the only person this sort of things happen to. Now I feel better :)
Permalink to Comment7. syn on August 3, 2006 9:26 AM writes...
If you guys were working for my PhD advisor, he'd "kill" you for that kind of unproducibility. I was made to feel like I was the only person this sort of things happen to. Now I feel better :)
Permalink to Comment8. Painful experience on August 3, 2006 9:34 AM writes...
It has to be one of the worst catch 22 situations you can be in as a grad student: 1st time important reaction gives 85%, 2nd time and on gives 5-25%. I spent 3 months a few years back trying to get a reaction to work. Eventually, we just ditched it and all the while my advisor makes me feel like I am an idiot!
Permalink to Comment9. Milo on August 3, 2006 10:43 AM writes...
My favorite comment from my advisor was: "Milo, this reaction must give a higher yield, try it again." Paper chemistry is always quantitative.
Permalink to Comment10. secret milkshake on August 3, 2006 3:47 PM writes...
Yea, I knw this knid of boss attitude. It is caused by overexposure to conference lectures and articles. Published/presented stuff is sold in tidied-up form and if your boss lives on such a diet too long without doing synthetic chemistry with his hands, he starts getting unrealistic.
Permalink to CommentIt is easy for such a boss to get a student-girlfriend that knows how to *realy* improve the yields.
11. MTK on August 3, 2006 5:35 PM writes...
One time in grad school I was trying to do a reaction for about a month. I was setting up 10 reactions at a time in vials, tlc-ing them, then tossing them when all I saw was about 10 spots. Finally, I had one reaction go to one big major spot. I isolated the product in ~85% yield.
I then spent the next two weeks trying to repeat it, only to see 10 frickin' spots every time. Of course, I did the obvious thing which was go back and scrupulously purify all my raw materials, buy new solvents, etc. Same stinkin' 10 spots.
I then figured well if an impurity isn't causing the 10 spots, maybe an impurity is causing the one spot. Went back to my notebook and noticed a different reaction I was running on the day of the success had a Cu catalyst. I reran my one 10 spot reaction, but added a speck of the Cu catalyst. Bingo! One spot. Best little piece of science I've done til this day. Told the story at a job interview and the interviewer almost offered me a job right there.
Feel free to use it.
Permalink to Comment12. Dustin James on August 3, 2006 6:18 PM writes...
Milo and others: a corrolary to Murphy's Law: "Nothing is impossible for the man who doesn't have to do it himself." Been known for a long time--my Dad, who was an Engineer at Enjay (before Exxon), had a copy of it in his desk.
Permalink to Comment13. syn on August 4, 2006 1:41 PM writes...
MTK,
Interesting. Care to share what reaction it was?
Permalink to Comment14. syn on August 4, 2006 1:42 PM writes...
MTK, I like you following comments, very insightful actually.
"I then figured well if an impurity isn't causing the 10 spots, maybe an impurity is causing the one spot. "
Permalink to Comment15. MTK on August 5, 2006 2:03 PM writes...
It was a Pd catalyzed coupling. The whole copper thing ended up being funny in that it helped in some cases and had no effect on other cases. As far as we could tell, there was an effect with electron deficient sp2 halides, but had no effect with others.
Now that I think about it, we never did run the control reaction without Pd. Hhhmmm.
Permalink to Comment16. Syn on August 7, 2006 12:45 PM writes...
Ah, CuTC is known to mediate Stille and Suzuki (Liebskin, Emory). Maybe you were seeing a similar thing?
Permalink to Comment