Wednesday, January 15, 2014

A Tale of Two Olefins


A prototype of the graphical abstract that Phil didn't want to submit to JACS.
Today marks the day that another paper makes its way out of the hot little presses of the ACS Publication Office and into the world of chemistry. This time, it's a paper from our lab that details a reductive coupling of an election-neutral olefin with an electron-deficient one. Since the chemistry is already described in the article, I wanted give an account of how this project came to be.

Apparently Phil's gotten at lot of flak regarding the two-phase approach to terpene total synthesis. To be frank, the oxidase phases of our projects get significantly more attention by the general synthetic community than the accompanying cyclase phases. Some of the people in our lab even choose to replace the cyclase phase with a "purchase" phase so they can get right into the meat of the oxidase phase. In my first months here, I was talking to Emily (my hoodmate) and she said something to the effect of "no matter how you look at it, C–H oxidations are sexy." 

Meanwhile, some would argue that the cyclase phases of our projects are simply comprised of Robinson annulations, conjugate additions, Diels-Alder cycloadditions, polyene cyclizations, ect.—"tried and true" reactions that your eye quickly passes over as you look for the unusual transformations in the synthesis. It would be pretty cool if we could breathe some extra pizazz into our cyclase phases and make them just as exciting as our oxidase phases.

Phil's musings on olefin cyclizations.
Phil outlined this all for me one day when I popped in his office to talk about new projects. At some point he sketched some structures that I've included above to help illustrate an idea he had to sexify our cyclase phases (I know there's a carbonyl and a methyl missing from the left- and rightmost structures, respectively, but I think we can forgive him). To the left is a cyclization of an enone that Snider was able to coax using EtAlCl₂ as a Lewis acid to give a bicyclo[4.3.0]nonane. The decalin to the right is a pretty recognizable motif in terpenes that Phil wanted to synthesize from the readily available starting material that Snider used. Like Snider's transformation, Phil wanted a single-step reaction where we wouldn't have to convert the olefins to other functional groups in order to achieve the cyclization.

I set out trying to achieve this transformation using conditions that were inspired by a paper from Boger's group and was later joined by Yuki-san (rock-star postdoc and basketball extraordinaire). We were able to extend the method to intermolecular couplings and the rest of the story is history, or rather, outlined in our paper. The end result is a reaction that's pretty easy to set up—it's literally a dump and stir reaction. I've even made a short video of it below for those who want a little extra convincing.


There's one last thing I wanted to mention involving the mechanism we proposed at the end of the paper. Although we outlined some mechanistic probes in the text of the paper, there were a couple of others that we did not include that appear to support the formation of radical intermediates. When the vinyl cyclopropane shown below was subjected to the reaction conditions, I was able to observe several products by GC/MS: starting material and another product with m/z = 144, four separate products with m/z = 146, and one product with m/z = 148. Some possible structures that fit these criteria are shown below. However, I was not able to isolate and individually characterize these products.

Vinylcyclopropane experiment and potential structures of the products.
It also would have been nice to obtain an EPR spectrum of a mixture of Fe(acac)₃ and PhSiH₃ to see if we could observe an Fe(III) hydride, but Scripps unfortunately does not possess an EPR spectrometer.

Anyways, if anyone has questions about the project or any of the background behind it, I'd be more than happy to answer them!

27 comments:

  1. kudos to you, for doing a 21 century chemistry. Back when I started to learn organic chemistry we were told that "radical reactions are ugly and hard to control."

    I noticed a Simax logo on your oil bath, from Czech glassware maker: How did that end up in California? And are they any good? The reason for asking is that the biggest crystallization dishes one can buy from Pyrex or Kimble-Kontess are 190mm wide (100mm high), so that barely fits a 2 L flask and they cost $60 apiece (or 300 USD for a case of 6). But the Kavalier guys are selling cryst dishes one size up, 230mm wide x 100mm, for about $30 apiece. So are they any good? - can you put them on a hotplate without cracking and spilling the oil all over?

    ReplyDelete
    Replies
    1. Thanks milkshake!

      And you have some sharp eyes! I'm not actually sure how we ended up with the Simax dish, but that particular oil bath has been taken up to at least 165 °C and there has been no breakage so far. It seems pretty durable.

      Delete
  2. for separating various olefins from saturated hydrocarbons, people have been running columns on silica impregnated with 10% AgNO3, it works because of transient complexation of Ag(+) with C=C. I have never done it myself but there were two things I remember: 1) low light environment, Al foil wrap, to prevent the darkening of the column 2) the hexane or petroleum-ether used was pre-treated by shaking with conc. sulfuric acid, to eat the traces of olefins, then filtered through pad of activated alumina.

    ReplyDelete
    Replies
    1. This is really interesting... essentially the first half of my Ph.D. was a nasty olefin problem, and everyone mentions the AgNO3 trick... yet I've never found anyone that had done it. I tried several procedures from stuff I found in the literature, and on the internet, and never had consistent enough results to really use it. I've done 1) the aluminum foil, but i've never seen 2) sulfuric acid. Got a reference for that? In case I ever need to revisit that chemistry?

      Delete
    2. I thought about doing that, so I talked to some people in the Shenvi lab who have done it. They told me that it works best with olefins conjugated to electron-withdrawing groups, so I'm not sure that it would have helped my case that much. However, I ran the vinylcyclopropane experiment as a curiosity after we submitted the paper and I felt like the GC/MS trace provided enough evidence to satisfy those curiosities.

      Delete
    3. Here's a few notes that I wrote down for anyone who might find them useful. These are probably elsewhere on the Internet, but:

      Preparation:
      Take a 10% solution of AgNO3 in MeCN and either run it up a prep plate or use it to make a SiO2 slurry. Rotovap the SiO2, dry it on the high vac, and then heat it in the oven.

      Notes:
      Excess moisture/MeCN on the SiO2 really messes up the separation. Having your impregnated SiO2 bone dry is critical.
      Do not expose to light, particularly when hot.
      You'll lose any sort of UV activity on your TLC plates.
      Don't use oxidizing stains like KMnO4.

      Delete
  3. Great post! I've also had occasion to use the Ag(+) olefin silica separation trick. The problem is the preparation of the gel: You have to slurry the SiO2 and AgNO3 together in water for a bit, then dry it in an oven until it flows like standard silica. Milkshake's foil and low light conditions apply. I was able to separate an endocyclic olefin from its saturated partner, no EWG needed.

    ReplyDelete
    Replies
    1. Thanks! Were you able by change to separate a tight pair of unactivated E/Z olefin isomers by any chance?

      Delete
    2. I think it has a chance to work because Ag(+) complex formation is how you can readily separate trans-cyclooctene after UV- irradiation of cyclooctene. But is a special case: the high strain C=C probably makes these Ag(+) complexes a lot more stable, the same like with cyclooctyne

      Delete
  4. Some questions from me:
    1. What about IR? The Fe-H stretch should be distinguishable from Si-H, no?
    2. Can the reaction work for poly-ene cyclization? eg. subject protected farnesol + MVK (or modified farnesol with an enone at the other end) to the reaction and what do you get?
    3. If C-H oxidations are sexy, what would be the C-H oxidation equivalent of twerking?

    ReplyDelete
    Replies
    1. 1. So I haven't tried IR monitoring and, to be honest, I was discouraged from using it when I read a review on Fe hydrides that said Fe–H stretching intensities are "often weak and hence the method is not entirely reliable."
      2. The only polyene cyclization I've tried so far is a cyclization of farnesal, which resulted in the ever feared complex mixture of products. I'm not sure if that was a substrate-specific thing or if all attempts at polyene cyclizations are doomed to fail.
      3. I can't think of a witty comment, so I'll leave that one to Emily :P

      Delete
    2. hmmmm...twerking is quite promiscuous...TFDO?

      Delete
  5. I have a proposal that is quite silly: Since acac ligand is quite similar in properties to Jacobsen's salen ligands, it would be worth trying the asymmetric version with [Fe(III)(salen)]OAc. It can be made simply by stirring Fe(OAc)2 with the inexpensive commercial optically pure salen ligand in ethanol+some toluene mix in an open beaker at room temperature for about one hour, then evaporating the solution.

    I realize that your reaction system is probably radical in nature, with Fe(III)-hydride species implicated in the formation of tertiary radicals. But Fe(II) and Fe(III) are fairly decent Lewis acid and likely coordinates to the carbonyl of the enone, as it is shown by the formation of the hydroxy hydrindane product (in the absence of a good silane), and it is there to reduce the addition-product alpha radical to enolate. So there is some chance for getting face selectivity on the system, without doing too much extra work apart from developing a chiral GC assay.

    ReplyDelete
    Replies
    1. We've considered trying to render the reaction asymmetric, but have never seriously tried (mostly because we didn't realize that the reaction could be run catalytically [Fe(acac)₃ is cheap enough to use stoichiometrically] until about a month before we prepared the manuscript—using ligands to control the enantioselectivity of the stoichiometric reaction would have cost big bucks). Your idea to use salen ligands is easy enough to try and is definitely worth a shot. Thanks for the suggestion!

      Delete
  6. Excellent work again!
    How much optimization/exploration went into finding the correct conditions/reagents? you show a little in the paper but i have to imagine it was non-trivial to get it right...

    keep up the good work, looking forward to the next release!

    ReplyDelete
    Replies
    1. Thanks for your kind words! We actually got really lucky with finding the final conditions. We started with trying to optimize the conditions using Fe₂(ox)₃ for about a month before we moved on—partially because we were getting side products, partially because iron oxalate is fairly pricey, partially because it took multiple hours to solublize in water—to using Fe(acac)₃. By that time we knew that PhSiH₃ was the reducing agent of choice, so from there it was only a matter of optimizing equivalents of everything. We went on carrying out the reaction stoichiometrically in iron for about 2 months before we realized that we could run the reaction catalytically (see above) and then we reran almost all of the reactions catalytically. Most of the yields were right around the same as the stoichiometric reactions, although there were some catalytic reactions that give slightly lower yields. Overall, I'd say optimization took around 2 months combined.

      Delete
  7. I just finished reading the paper and the supplementary material, and looked at the video.
    Is this reaction exothermic? You use 60°C after throwing everything together, so I would be a bit worried about radicals and accumulation of reactive intermediates causing some thermal problems on scale-up. By scale up I mean >100g.
    What is the largest scale you have done?

    ReplyDelete
    Replies
    1. Quintus, I have not noticed an exotherm upon adding the PhSiH₃ nor have I noticed one 5-10ish minutes after the addition of PhSiH₃ prior to heating in the oil bath. If there is one, perhaps it's only apparent on very large scale reactions (by academic standards). The largest scale we have run this reaction on is on a gram—we've only had a limited amount of PhSiH₃ throughout this project (for some reason, only one 5 gram bottle of the multiple bottles of PhSiH₃ that we've ordered has arrived) and we could not spare enough for say a 10 gram reaction.

      I gather you work in process—if you ever get the chance perform this reaction on >100 grams, we (especially Phil) would be very interested to know the outcome.

      Delete
  8. Thanks for the reply. I would like to try it out but it's finding the time. I also noticed that your reaction is very very dilute, any reason for this?
    I was a bit disappointed to see you just threw everything together so there would be a significant chance of accumulation happening, perhaps leading to unwanted thermal events. But on 20mg scale I suppose that's not important. However, that said, if would have made a nice addition to the paper if such details could be included. I don't expect process research at all, that's not your job, but an indication as to which way the wind blows is always useful for others.
    I wish you continued good luck and many more papers.

    ReplyDelete
    Replies
    1. Thanks Quintus! The intramolecular cyclizations were all dilute because we initially thought that we might get some competitive polymerization. Since then, I have not noticed any difference between running the reactions at 0.05–0.5 M using the same substrates—I just went for the more dilute reactions because I felt more comfortable with them. I have not tried anything more concentrated than 0.5 M though. We have also not tried the intermolecular reactions more concentrated than what was written in the SI.

      During the first few reactions I tried, I did not notice the order of addition of the reagents/solvents having any significant effects on the outcome of the reaction (I always added the PhSiH₃ last though). I just threw everything together and rinsed down the sides of the culture tube with the solvent in the video/SI because that was the fastest and most convenient way for me. You could probably add the catalyst to a solution of your substrate in EtOH/ethylene glycol, followed by addition of the PhSiH₃ without any negative consequences.

      If you ever do find the time to test this out on a larger scale, please let us know!

      Delete
  9. Thanks for the additional information. I shall certainly get in touch if I get a chance to do this chemistry although with the current projects I can't see it fitting in atm. Perhaps later!
    best to all
    Quintus

    ReplyDelete
  10. Nice paper Jlo :) good luck for the futur

    ReplyDelete
  11. Very nice paper!

    I'm wondering whether your conditions can be used with alkyne electrophiles (ethyl or methyl butynoate), or whether they can tolerate alkynes in general (do they reduce or undergo fe-catalyzed hydrosilation)?

    ReplyDelete
    Replies
    1. Thank you! We've actually tried using ethyl butynoate as an acceptor and unfortunately did not observe any coupling. We've been meaning to explore alkynes as the donor component, but we haven't had time to try yet.

      Delete
    2. IMHO terminal allenes would be the really interesting ones. Some mono-subst allenes are easily made with propargyl bromide

      Delete