Weix wins with hindered aryl halides

 Citation:

Wu, T.; Castro, A. J.; Ganguli, K; Rotella, M. E.; Ye, N.; Gallou, F.; Wu, B.; Weix, D.J. Cross-Electrophile Coupling to Form Sterically Hindered C(sp2)-C(sp3) Bonds: Ni and Co Afford Complementary Reactivity. J. Am. Chem. Soc. ASAP. 

https://doi.org/10.1021/jacs.4c16912

 Summary Figure:

Background:

If you look at a random cross-coupling paper with aryl halides, one thing you might not notice (until you go to plan a substrate scope) is that usually substituents are para to the halide. As an example, I'll show the substrate scope of this great Ni/Co co-catalysis paper from Xue and Zheng (this was one of the other papers I was considering writing on this week):

 

(https://pubs.acs.org/doi/10.1021/jacsau.5c00031)

That's a lot of substrates! However, if you look at 1-32 (the aryl halide scope), only 2 of them have an ortho substituent: 27 and 21. 27 barely counts, as it's a 1-naphthyl, so really only 21 is an ortho-substitued aryl iodide, and it only provided 27% yield. 

First, it's pretty easy to make para-substituted compounds. They give nice NMR spectra because the aromatic peaks have some symmetry. There are lots of easily available building blocks (see 1-15), and known ways to convert them into interesting compounds. For example, 31 was made by taking 4-iodoaniline and doing an amide-bond forming reaction. However, ortho-substituted aryl halides building blocks are also fairly easy to find. On Sigma-Aldrich, 4-iodoaniline is 25 g for $60.80 while 2-iodoaniline is 5 g for $51.70. Not a substantial difference. 

However, a lot of the time ortho-substituted compounds don't work. It usually isn't steric hindrance blocking oxidative addition, as that's how you usually make stable oxidative addition complexes, but the problem ends up being reductive elimination or migratory insertion.

 

(For example, see these Doyle papers on nickel complexes: https://pubs.acs.org/doi/10.1021/jacs.0c00781, https://pubs.acs.org/doi/10.1021/jacs.7b13281)

If you let an Ar-Ni(II)-X complex sit around for too long, it's likely to produce biphenyl dimers via transmetalation/disproportionation and then reductive elimination. That's a long-known reaction (see Kochi: https://pubs.acs.org/doi/pdf/10.1021/ja00519a015). The ortho-substituent stabilizes the complex by preventing reductive elimination via steric hindrance.

However, preventing reductive elimination is not good when you're trying to form a C-C bond. This property makes it difficult (although not impossible) for these substrates to actually participate in cross-coupling. If you're a methods chemist who just developed a new reaction and wants to show off the scope, why test 2 variables by attaching a questionable functional group at the ortho-position when you could just do it para- and not worry? This leads to the entire field incorporating a hidden bias against ortho- substituents.  

The goal of this paper by Weix and coworkers was to specifically target ortho-substituted aryl halides and identify reaction parameters that could work on them- hopefully discovering either new classes of ligands or additives or reaction conditions that can be added to the bag of tricks for troubleshooting these reactions. 

 How it works:

The Weix lab took a general cross-electrophile coupling (XEC) reaction between an aryl bromide and a secondary alkyl bromide and put a bulky group (either isopropyl or dimethylamine) at the ortho-position. They then tried a couple of known literature conditions for XEC reactions between aryl bromides and alkyl halides, each of which gave product in somewhere between 30-60% yield, which isn't bad for what is expected to be a difficult reaction. A little more optimization (changing up the bipyridine ligand substituents, a better nickel precatalyst, taking out some of the unnecessary additives, increasing the temperature and time) got the yield up to 76%. 

 

 They then did the same thing for cobalt, which can also do similar reactions. Here they found that while nickel can't do doubly-substituted aryl halides, cobalt can!

 

 So then the question is, why don't reactions work? What's the failure mode? The authors identified that a lot of the mass balance was going to reduction products:

 

This is also confirmed by some stoichiometric studies, including:

 

So the authors propose 3 routes for the formation of these reduction products. 

First, the formation of an organometallic reagent, which is quenched upon workup. 

 

The formation of organozinc intermediates via nickel or cobalt catalysis is known. 

The common way to check for this is to quench the reaction with an electrophile that gives some sort of detection. For example, if you quench with I2 you get the aryl iodide, if you quench with D2O or CD3OD you get deuterium incorporation. In this case, the authors quenched the reaction with deuterium and saw <2% deuterium incorporation, which is consistent with the hydrogen not coming from the quench. Instead, it must have come from something in the reaction mix, which given the lack of ready H+ sources, means it is unlikely to be from an organometallic intermediate. Note that this does not mean impossible, it just means that if there is another reasonable explanation, that explanation jumps ahead in line. 

Second, the hydrogen could have come from beta-hydride elimination from the alkyl bromide:

 

Basically, after you've gotten both electrophiles onto the nickel catalyst via oxidative addition steps, you could either reductively eliminate to form the desired product, or do beta-hydride elimination to form a nickel hydride. If you do reductive elimination after that, you can form a C-H bond. I've drawn it as a hydrogen or deuterium, because one of the ways to check for this is to use a fully deuterated alkyl bromide. If you see deuterium incorporation into the reduced product, it means that deuterium came from B-H elimination. This is what the authors see: when d7-iPr bromide was used as the alkyl electrophile, 47% of the reduced product had a deuterium at that position, which is consistent with B-H elimination being the off-cycle pathway.

 The third possibility is the formation of an aryl radical, which can abstract a hydrogen atom (both the proton and an electron) from another molecule, which is usually the solvent, since it is in excess. Both DMA and THF have abstractable protons, as it the case with most solvents, especially because aryl radicals are very unstable. This aryl radical can form either as part of the oxidative addition:

 

Or by metal-carbon bond homolysis: 

 

As these processes are related, it's quite difficult to differentiate them. 

The way to test for this method of forming the reduction product is to use the deuterated solvent. If you use d8-THF (because you're rich) or d3-toluene (if you're poor) or another deuterated solvent (d9-DMA if if you're the authors) and observe deuterium incorporation, it means that the reduction is coming from the solvent. This is likely via H-atom transfer, as these positions are not acidic but the radicals are more stable than aryl radicals.

Interestingly, with the di-ortho-substituted nickel system (which gives very low desired product and mostly reduction byproduct), the test for beta-hydride elimination was negative. The test for aryl radicals was positive: with d9-DMA, 49% deuterium incorporation was observed.  I think this is very cool, and shines light on why these substrates seem to fail for the nickel system- a new off-cycle pathway becomes predominant. 

The cobalt reactions were interesting: quenching with D2O did not observe deuterium (not forming organometallic reagents), using iPrBr gave only 14% deuterium incorporation (maybe B-H elimination is operative), and using deuterated DMA or THF gave 24 and 20% deuterium incorporation, respectively. These are not high numbers, which makes this result somewhat inconclusive. 

The authors then turned to DFT to see if that could provide answers, which it really didn't. The one-electron oxidative addition pathway (which forms aryl radicals) for cobalt was 16 kCal/mol higher than the two-electron pathway that doesn't make aryl radicals. That number is pretty high, and suggests it isn't feasible. The bond dissociation energy of the cobalt-carbon bond is ~20-30 kCal/mol, which also seems difficult to break willy-nilly. 

Initial Questions and Key findings:

1. Can you develop a set of conditions to do cross-coupling reactions with ortho-substituted aryl halides?

A: Yes, existing conditions can get okay yields, some optimization can make it great yields. There are differences between the nickel and cobalt reactivity that can be exploited to activate many different substrates. 

2. Why is it so hard to do cross-coupling reactions with ortho-substituted aryl halides?

A: You begin to see reduction products forming due to off-cycle pathways including beta-hydride elimination from the alkyl coupling partner and the formation of aryl radicals. Identifying ways of suppressing these side pathways can enable improved reactivity.  

Takeaways:

There are a lot of ways reactions can go wrong, and when a reaction fails, it can be helpful to look at the byproducts formed and try to imagine how it got there. This is how you begin to identify new modes of reactivity and generate new reactions.