Ace in the Hole

Figure 1. Taken from Ref. [2].

On the second day of the Como Fluorine Conference, Prof. Makato Fujita has given an impressive talk on supramolecular chemistry and molecular recognition involving fluorous compounds. How to make ‘chemical nemesis’ to become pals? You have to see this, mate!

Figure 2. Blue dotted lines represent halogen bonding.

Let’s take a closer look at halogen bonding involving heavily fluorinated compounds, as the fluorine atoms are certainly not in the most low-key presence here (Figure 2). In a sense, the electron withdrawing nature of fluorine atoms induces halogen bonding. The nature of the iodine atom is also important. First, the electron deficiency on the iodine atoms on the perfluoroalkyl compounds is caused by large electron deformation of the soft lone-pair electron on the iodine atom, which is in turn induced by electron-withdrawal through sigma bonds. That will lead to a strong interaction with the Lewis basic component.  On the other hand, the fact that halogen bonding is strongest in iodides may point to the fact the softness of the lone-pair electron on the iodine atom is likely to play an essential role in this intermolecular interaction.

How can we observe a halogen bonding experimentally? Well, we can do NMR (vide infra), and after complexation of the Lewis base and the halogen donor, a upfield shift of the CF₂ signal in the 19F NMR can be observed. What is even better, is that the complexation through halogen bonding can induce liquid crystallinity from non-mesomorphic components.In many cases, a crystalline 1:1 adduct can be afforded, and its structure can be confirmed by X-ray crystallographic analysis. So, we have an elegant approach to verify halogen bonding – don’t you find crystals beautiful?

Figure 3. Taken from Ref. [1].

Prof. Fujita and his group wants to achieve a ‘fluorous recognition’ in synthetic cavities, such as molecular cages. The idea is to form a host-guest complex, where the ‘guest’ this time is of course the fluorous compound. Traditionally, there are 2 methodologies to do so – solvent size exclusion (driving force is C_xF_y group) or through hydrophobic effects (like C_xF_yCOO^–Na⁺. Indeed, they have published in Science Magazine in 2006 (Figure 3), where they have found a host-guest complex by putting 24 ligands to 12 metals (Pd²⁺). What is spectacular is that the complex, which contains non-polar fluorous groups, can be dissolved in the polar NMR solvent d6-DMSO and D₂O! That means the fluorous tails are protected by the host cage, which can be solvated in a polar solvent. Once again, by looking at the structure of the ligand, we can see the importance of having a (CH₂)_n spacer in the fluorous compound. The ‘spacer’ idea is a common theme in so many of the talks in the conference, from ligand design to fluorous membranes to biological inhibitors. Except if you want all the electron-withdrawing fluorine atoms to cast a huge impact on your ligand core, you have to put these spacers in between the fluorous groups and the ligand core, and hopefully, by careful design, it will still achieve a reasonable fluorine-phase solubility.

 Then, Prof. Fujita proposed 2 new and interesting strategies to achieve his aim – fluorous aggregation and halogen bond (XB)-assisted fluorous recognition. For the first concept, consider the following 2 equilibriums (Figure 4).

Figure 4. Taken from Ref. [2].

We can easily see that, when the host:RF = 1 : 1, the equilibrium is in favour of the direction where the fluorous compound is NOT in the cage. However, if a number of fluorous compounds are in the system, then due to the like-dissolve-like principle, the fluorous compounds will aggregate and they will stay inside the host cage (Figure 5). Indeed, a number of fluorous compounds are capable of doing this, and what is even more fascinating is that the system is showing a phenomenon molecular biologist should be familiar – co-operativity. Just like the case of hemoglobin, when multiple fluorous compounds enter the case, the binding shows the classic, sigmoidal Hill co-operativity. Different cage : fluorous-compound complex shows different stiochiometries. For example, a perfluorocycloalkane C₅F₇H₃ , gives a 1:4 adduct; while a linear HO-CH₂-(CF₂)_n-CH₂-OH gives a 1:2 adduct. All these have been verified by 19F NMR studies, and the complexes can be crystallized and subject to X-ray analysis, affirming the complex formation.

Figure 5. Taken from Ref. [2].

For the other strategy, Fujita et. al. has also employed halogen bonding to meet the challenge (Figure 6). From Figure 2, we can see the halogen bonding network in between the nitrogen and iodine atoms, and the establishment of the sigma hole (also in Prof. Diederich lecture).

Figure 6. Taken from Ref. [3].

While the space in the molecular cage is confined, the water molecules or counteranion NO3– can form halogen bonding with the iodine atoms on the fluorous compounds, and this has been verified by 19F NMR titration experiments and also X-ray crystallography. When aromatic compounds C₆F₃I₃ and C₆H₅NMe₂ are mixed, a donor-acceptor pair can be formed in the confined cavity, through the N---I halogen bonding (Figure 7). 

Figure 7. Taken from Ref. [3].

Thus, even the worst of enemies can become the best of friends – at least in a chemical sense.

by Ed Law
14/9/2015

Reference:

1. Science 2006, 313, 1273.

2. J. Am. Chem. Soc. 2014, 136, 1786.

3. Angew. Chem. Int. Ed. 2015, 54, 8411.

Le Trou (The 'Sigma' Hole)

Figure 1

I would like to talk about a number of great topics I have learnt from the Como Fluorine Conference in August. 2 of the keynote lectures, by Profs. François Diederich and Makato Fujita respectively, shed light on halogen bonding and its potential applications in molecular recognition. This time I will talk about Prof. Diederich’s one, I will leave Prof. Fujita for next time, as both talks are extremely impressive.

Prof. Diederich and his group has done a lot of impressive research in the field of molecular recognition, and some of the key intermolecular interactions include orthogonal dipolar interaction, halogen bonding, and pi-stacking.

An orthogonal dipolar interaction concerns an interaction of the type C-X ----- C=O. In one of the cases, it is a C-F ----- C=O interaction. For example, when the molecule nilotinib is interacting with the DFG loop of a protein, the fluorine atom on the CF₃ group can interact with the C=O bond on a particular amide group in the protein, in an intermolecular and orthogonal manner (Figure 2).

Figure 2

The other fascinating aspect of their research is the halogen bonding interaction (Figure 3). Halogen bonding (XB) can be defined as a non-covalent, intermolecular interaction of a Lewis basic group and the electrophilic part of a halogen group.  The term ‘σ hole’ is often termed and it is described as the area where the lone pair of the Lewis base is aligned with the σ* orbital of the carbon-halogen bond. Indeed, the size of the σ hole increases with a lower hybridization (sp) of halogen-bond donor. In 2014, Diederich et. al. has synthesized a number of (iodoethynyl)benzene derivatives, and study their halogen bonding properties with the Lewis basic quinuclidine. From the linear Hammett analysis, they have concluded that the para-substituent on the aromatic ring is in electronic communication with the iodethynyl group, and this can affect the halogen bonding interaction of the given compound (Figure 4).

Figure 3. The blue dotted-lines represent halogen bonding.

Figure 4

Thereafter, they have synthesized a 2-component, donor-acceptor type supramolecular architecture. The ‘donor’ side consists of 4 lutidine rings, where the nitrogen atom is Lewis basic (Figure 5). On the other hand, the acceptor side consists of 4 tetrafluoro-iodophenyl groups, and the iodine atom, closest to the donor side, is the ‘halogen donor’. Thus a N --- I halogen bonding appears, and the group has used 19F NMR binding titration studies and 2D NMR to show this has taken place.  A 1H, 19F HOESY NMR experiment has been used to map out the intermolecular H-F coupling (Figure 5). The cross signal in the 2D spectrum showed the through space interaction, and only the fluorine atoms ortho- to the iodine atom could couple through space to the hydrogens on the lutidine core. Along with the other data, the researchers have been able to verify the halogen bonding in action.

Figure 5.

When looking at single halogen bonding in solution, studies show that it is largely enthalpy driven, and also have a large unfavourable entropy at the same time. It also seems evident from the studies that, for a given, general structure of an inhibitor, a different substitution on the aromatic ring can drastically alter the binding constant K. The key explanation is because in some cases, an enthalpically frustrated water can be positioned in between the interaction sites, and in other case it is displaced. Thus, to improve inhibitory effects, a systematic understanding (which will involve a lot of modeling) has to be achieved, and it is great to see that many of these insights have already emerged from the research literature.

by Ed Law

6/9/2015

Reference for this blog article from:

Angew. Chem. Int. Ed. 2015, 54, 1.

Angew. Chem. Int. Ed. 2015, 54, 3290. A review.

Org. Lett. 2014, 16, 4722.

Phys.Chem. Chem. Phys., 2013, 15, 11178. A review.

ChemMedChem 2011, 6, 2048.

Science 2007, 317, 1881.

ChemMedChem 2006, 1, 611.

Angew. Chem. Int. Ed. 2004, 43, 5056.

ChemBioChem 2004, 5, 666.

Angew. Chem. 2003, 115, 2611.

SteadiCHEM (Fluorous Zinc Reagent)

At Expo Milano 2015, inside the 'Russia World'. Taken by Ed Law.

In the Como fluorine conference, I had a stimulating discussion with an insightful co-worker, who worked in the synthesis of ‘fluorous buckybowls’. He has pointed me to a great paper by Prof. Mikami et. al. in Chemistry – an European Journal earlier this year. I have read it and it is impressive – I would love to talk about that here!

Copper Catalyzed coupling reaction of perfluoroalkylzinc reagent. Adapted from [1].

Prof. Mikami’s group has reported the synthesis of a perfluoroalkylzinc reagent, of the structure (R_F)₂Zn(DMPU)₂, and its application in a copper-catalyzed coupling reaction, that results in perfluoroalkyl aromatic / heteroaromatic compounds. [1] The zinc reagent itself, unlike many main group organometallics involving Li and Mg, is a stable white powder. Yet, as the title of the article has pointed out, it is a reactive reagent. Like Charlie Bronson, he may have the leisure to play a harmonica, but he can shoot, too.

A trifluoromethylation reaction. Adapted from [2].

Some time ago, they have reported a trifluoromethylation reaction of aromatic halide using a copper catalyst. [2] The perfluoroalkylzinc reagent is generated in situ from CF₃I and zinc dust. In order to improve the versatility of the methodology, they have then developed the new reagent, (R_F)₂Zn(DMPU)_2.

Generation of the stable perfluoroalkylzinc reagent. Adapted from [1].

The perfluoroalkylzinc reagent is generated by a reaction between perfluoroalkyl iodide (R_F-I), diethylzinc, and DMPU at -60^oC, using hexane as a solvent. After warming to -20^oC and stirring at this temperature for 48 hours, the reaction mixture is worked up and finally they afford an air-stable white powder (except when R_F = CF₃, which is stable under argon).

The coupling reaction can be done in one-pot. After all the reactants (perfluoroalkylzinc reagent, copper salt, and aromatic halide / vinylic halide...) are loaded into the flask at room temperature under argon, the reaction takes place at elevated temperatures, affording the coupled product after workup. They have been able to couple 5 different perfluoroalkyl groups to 2 types of coupling partners – a substituted aromatic halide or a vinylic halide.

A selection of achieved chemical products. Adapted from [1].

The most important step in the reaction pathway is the transmetallation of the perfluoroalkyl group, from zinc to copper center. The researchers have carried out nice 19F NMR experiments to elucidate the possible intermediates, and these results have proved to show important insights regarding the reactivity of the different types of perfluoroalkylzinc reagents. Before I go into these, it would be more logical to look at the observations first.

Though belonging to the fluorine universe, the results in a trifluoromethyl (CF₃) group is rather different from the longer perfluoroalkyl chains (C₂F₅, C₃F₇ , C₆F₁₃ ...). In most cases, a 10 mol% loading of the copper catalyst, CuI, is enough to effect the coupling reaction at elevated temperature, and in some cases even a lower catalyst loading may also work. For the longer perfluoroalkyl chains, the electronic status of the aromatic compound (whether it contains an electron withdrawing or electron donating substituent) is not important, although electron donating groups lead to a longer reaction time. This is NOT the case for the CF₃ group. While aromatic rings containing electron-withdrawing groups work well, no reaction can be observed for those substrates with electron-donating groups (methoxy in this case), even when a stoichiometric amount of CuI is used. When employing 1 eq. of another copper salt, CuTC,   successful reaction with a good yield can be afforded. This observation indeed confirms what the researchers have discovered in the 19F NMR experiments. When they mix (CF₃)₂Zn(DMPU)₂ with CuI, they disover 2 singals in the 19F NMR specturm, which correspond to 2 different anionic copper species. When they mix the same zinc reagent with CuTC,  another 2 different signals appear, and one of them is likely to corresponf to CuCF₃. Indeed, CuCF₃ is a likely intermediate in many of the copper-catalyzed coupling reaction involving trifluoromethyl groups, and these observations may imply that distinct copper intermediates are involved in all these coupling reactions.

It can also be seen that, the longer reaction time regarding the long perfluoroalkyl chains is likely attributed to a slower zinc-to-copper transmetallation step. And, the temperature range (50-120 ^oC) also shows the thermal stability of the perfluoroalkylcopper intermediates in these reactions.

They have also proposed a possible mechanism.

Proposed reaction mechanism. Taken from [1].

This is an impressive work. In the Como Conference, I have the great opportunity to meet Professor Kenji Uneyama, who has done a lot of research on organometallics involving fluorinated compounds. In his wonderful lecture, he has shown that while a lot of organometallics can be made (Li, Mg, Cu, Zn), many of them can be rather unstable and can only be used at very low temperature (including cuprates of the types R₂CuLi or R₂Cu(CN)Li₂). And, it can be hard to predict (or have strong confidence) which structure will generate a stable one, you have to experiment on it! I have the impression that those organometallics involving Cu and Zn tend to be more stable at elevated temperature, and therefore should find wider applications in coupling reactions. Given the emergence of many of these great perfluoroalkyl reagents (see also my recent article ‘The Magnficent ATEbersons’, R_FZn(Me)Cl^-Li⁺), this certainly is a burgeoning field.

By Ed Law

1/9/2015

Reference:

1. Stable but Reactive Perfluoroalkylzinc Reagents: Application in Ligand-Free Copper-Catalyzed Perfluoroalkylation of Aryl Iodides

Kohsuke Aikawa, Yuzo Nakamura, Yuki Yokota, Wataru Toya, and Koichi Mikami

Chem. Eur. J. 2015, 21, 96 – 100

DOI: 10.1002/chem.201405677

2. Y. Nakamura, M. Fujiu, T. Murase, Y. Itoh, H. Serizawa, K. Aikawa, K. Mikami, Beilstein J. Org. Chem. 2013, 9, 2404.

Added on 2/9/2015:

I think I have missed a few interesting aspects in yesterday's article. First, a bit more on the Zinc / copper perfluoroalkyls. Zinc and copper fluorous organometallics are more thermally stable than their Li / Mg counterparts, due to a stabilization from a more 'covalent' character in the metal-carbon bond, which is originated from a softer nature of the metal. Thus, these compounds tend to be able to survive a higher reaction temperature.

Second, an observation in the reaction is that, when electron-donating group substituted aromatic halides are used as reactants, 15% of pentafluoroethylated by-product can be observed, when 1 eq. of CuTC is used at 50^oC. How does this side product arise? Well, there is evidence that when trifluoromethylcopper (I), CF₃Cu is generated,  CF₃Cu is in an equilibrium with [CF₂Cu]F^-. This can be built up to a longer perfluoroalkyl copper chain, through a carbene insertion mechanism. This 'oligomerization' can be stopped by adding HMPA. The carbene insertion equilibrium is the likely reason for the formation of perfluoroethyl side-products in this coupling reaction, and it has also been observed in other copper systems. The pentafluoroethyl side-product can be rather difficult to remove from the crude mixture. As stated, this side-reaction can be counteracted by adding HMPA (the researchers have used DMPU here instead) or lower the reaction temperature. Yet,  I suppose 50^oC is already the lowest temperature the researchers can get to, and have ca. 75% isolated yield for the product is already impressive.

by Ed Law

2/9/2015