Download PDF
Synlett 2021; 32(05): 447-456
DOI: 10.1055/s-0040-1706407
account
The Power of Transition Metals: An Unending Well-Spring of New Reactivity
© Georg Thieme Verlag Stuttgart · New York

From Enantioselective to Regiodivergent Epoxide Opening and Radical Arylation – Useful or Just Interesting?

Andreas Gansäuer∗

This work was supported by the Deutsche Forschungsgemeinschaft (DFG) (Ga 619/12-1).
Further Information

Publication History

Received: 07 May 2020

Accepted after revision: 02 June 2020

Publication Date:
20 July 2020 (online)

 
 PDF Download Permissions and Reprints All articles of this category


Abstract

It is shown how the simple titanocene(III)-catalyzed reductive opening of epoxides has developed from a highly selective desymmetrizing opening of meso-epoxides. The combination of the origins of regioselectivity with atom-economic catalysis in single electron steps results in a highly selective regiodivergent epoxide opening (REO).

1 Introduction

2 Regiodivergent Epoxide Opening

3 Atom Economy and Epoxide Arylation

4 REO Arylation

5 Words about the Wording

6 Conclusion


#

Key words

arylation - catalysis - diversity-oriented synthesis - epoxides - regiodivergent - titanocenes

Biographical Sketch

Zoom Image

Andreas Gansäuer was born in Paris and studied chemistry at the University of Bonn and the University of Oxford. He obtained his Ph.D. in the group of Prof. Reetz at the Max-Planck-Institut für Kohlenforschung. After a postdoctoral position in the group of Prof. Trost at Stanford University, he completed his habilitation with Prof. Brückner in Göttingen. After a short time in Freiburg, he became professor of organic chemistry at the University of Bonn in 2000 and has remained there ever since. His research interests include radical chemistry and catalysis.

1

Introduction

Sometime after the paper on the regiodivergent arylation of epoxides by Felix Mühlhaus, Hendrik Weißbarth and myself appeared in press,[1] Prof. Peter Vollhardt (University of California, Berkeley) sent me an e-mail asking for an ‘encore account’ on the concepts of this work (and how they came about in reality). The invitation was a very pleasant surprise and I decided to accept it quickly for the following reasons. First, the concepts finally combined in this paper, regiodivergent epoxide opening and atom-economical radical arylation, have been the focus of my group for some time. Second, such an account is an appropriate medium for personally applauding the important contributions of the many young scientists who dared to join my group, not knowing what to expect and who made this pleasant success possible. Finally, the first account was a big boost in morale right at the start of my independent career and writing the ‘encore account’ is a way to say thank you to Synlett for that very important initial support.

The appropriate starting point for the story is the concept of regiodivergent epoxide opening (REO). This is also (but not only) so because the section on epoxides in the first account ended with the optimistic statement that ‘asymmetric versions of the reaction are also being investigated in our labs in Göttingen and first results are very promising, indeed’.[2] I was trying to bolster myself up. Let us see what happened.


# 2

Regiodivergent Epoxide Opening

When starting an independent career, choosing the ‘right’ area of research is the most important issue. My Ph.D. supervisor, Manfred Reetz, gave me the valuable and unexpected advice to do something that neither he nor Barry Trost (my postdoctoral supervisor) were famous for. Neither of the two had done radical chemistry, so the general area of research was clear. Enantioselective catalysis and organometallic chemistry were my own favorite topics and so metal-catalyzed enantioselective radical chemistry just had to be my field. To me, Nugent and RajaBabu’s titanocene(III)-mediated epoxide opening[3] seemed an ideal starting point. The metal was essential for radical generation and even simple C5H5 ligands imposed impressive control on radical formation and the ensuing radical reactions. My first success as an independent researcher was to make the reaction catalytic.[4] How this was achieved is described in detail in the first account.[2]

In brief, the stoichiometric amount of Cp2TiCl2 was replaced by a catalytic amount of Cp2TiCl2 (5–10 mol%) and a stoichiometric amount of Coll·HCl (2,4,6-collidinium chloride). Of course, this set the stage for enantioselective catalysis. Inspired by Jacobsen’s great successes,[5] Thorsten Lauterbach, Harald Bluhm and I, working on epoxide opening at the time, were also very confident of being successful in the desymmetrization of meso-epoxides.

Of course, the critical issue in the development of a new enantioselective catalytic reaction is the choice of the catalysts’ ligands. There were no related examples, and so we were clueless (in every sense of the word) of where to start. It was clear that Brintzinger’s complex 1 [6] (Scheme [1]) simply had to be tried due to its many successful applications in synthesis. However, we were slightly skeptical about its enantioselectivity with respect to epoxide opening. The ketone and imine substrates of the highly enantioselective hydrogenations catalyzed by 1 certainly bind to titanium differently than an epoxide does.[7] The fact, that I had studied the chemistry of titanocene complexes purely out of curiosity during my Ph.D. in Mühlheim turned out to be a great help in choosing our second candidate, Vollhardt and Halterman’s titanocene 2.[8] It features ligands derived from phenylmenthol. Phenylmenthol had been successfully used as a chiral auxiliary. π-Stacking of the phenyl group with the covalently bound[9] substrate had been proposed as a control element. Based on simple models, we hoped that a similar interaction between the cyclopentadienyl ring and the phenyl group would also ‘lock’ the conformation of 2.

Due to inferior results with menthol-derived auxiliaries, and bearing in mind that no X-ray structure of 3 was available at the time, we had assumed that such a mechanism of ‘conformational locking’ was not available for Kagan’s complex 3 that features menthyl-substituted ligands lacking the phenyl group.[10] Using a C 2-symmetrical catalyst for differentiating the substituents of a meso-compound seemed a good idea. One of the substituents has to interact with the ligands of the catalyst while the other one has to point into empty space. Such differentiation just had to enforce a high selectivity for ring opening.

Zoom Image
Scheme 1 Structures of the catalysts investigated in the desymmetrization of meso-epoxides

As substrates we chose epoxide 4 derived from (Z)-2-butene-1,4-diol for reduction reactions and cyclopentene, cyclohexene, and cyclohepteneoxide for 1,4-addition reactions to acrylates.[11] Some of the results are summarized in Scheme [1], showing that 1 is indeed not well suited for enantioselective epoxide opening. However, complex 2 gives excellent results in the opening of epoxide 4, while the selectivity is reasonable with cyclopentene oxide 5. The dia­stereoselectivity of radical addition to the acrylates with 2 is noticeably higher than with Cp2TiCl2. Therefore, both radical generation and radical trapping were reagent controlled. We had done it; our catalytic enantioselective radical reaction was working!

With hindsight, it seems that Harald, Thorsten and I had some element of luck with our choice of the catalysts. After all, we had had a clear and seemingly correct rationale for the catalyst design. However, from a practical point of view, catalyst 2 turned out to be a tough choice since it is not very easy to make on gram scale. One tricky chromatographic separation was especially tedious on large scale. In the end, Thorsten and Harald made a wise decision. They decided to use the much easier to prepare catalyst 3. Their courage paid off! Catalyst 3 was at least as selective and as active as 2.[11]

After obtaining the X-ray structure of 3 we understood why (Figure [1]).[12] The conformations of 2 and 3 are essentially identical with the Ph group in 2 being replaced by a H atom in 3. The ‘flat’ Ph group must be considered as smaller than a ‘three-dimensional’ CH3 group. This insight was slightly embarrassing for me. It was the essence of a number of comments on B-Lewis acids that Manfred Reetz had made to me during my Ph.D.

Zoom Image
Figure 1 X-ray structures of 2 and 3 [12]

So, we had in hand an enantioselective opening of meso-epoxides, being a desymmetrization of the enantiotopic C–O bonds of the substrate by a radical mechanism. The identity of the substituents of an epoxide that cause the enantiotopicity and that make the substrate a meso-compound is a fundamental limitation for the generality and synthetic usefulness of the approach. So, the reaction was probably interesting, but certainly not that useful. That did not bother me too much, though. The reaction got me my position in Bonn, a great city, where I essentially grew up and have stayed ever since without any regrets!

Moving to Bonn was a temporary halt to enantioselective catalysis in my group. We had to settle down, new group members had to be recruited, courses had to be designed, and most importantly, we had to integrate into the department’s research structure. In the German system, this can be a big effort, and in Bonn it, luckily, was. The Corporative Research Center (CRC) 624 on Templates was founded in 2002 and eventually provided funding for 11 years. The CRC was not focused on catalysis or radicals. Rather, it was dealing with the structural organization of objects by non-covalent interactions. My group contributed to a number of topics such as selectivity in cyclizations,[13] aggregation phenomena via non-covalent interactions,[14] and interactions of titanocenes with solvents.[15]

However, the distraction, as pleasant as it was, did not last too long. Together with Stefan Grimme, at the time still in Münster, and Kim Daasbjerg in Aarhus, we started to work on the mechanism of epoxide opening by Cp2TiCl. As one highlight of these studies (Figure [2]), Christian Mück-Lichtenfeld calculated the structure of the complex of 3 [in the oxidation state (III)] with 4 (3·4).[16]

The reasons for the selectivity of ring opening became obvious and confirmed the simple hypothesis of my very first grant application. One of the substituents of 4 is ‘below’ one of the ligands and the other substituent points into empty space. Since both radicals formed are exposed to identical electronic effects (a virtue of the identity of the meso-epoxide’s substituents), only the difference in steric interactions can be responsible for the selectivity of ring opening.

Zoom Image
Figure 2 Structure (left) and schematic representation (right) of the catalyst–substrate complex 3·4

This selectivity originates from releasing the unfavorable interaction of the ligands with the epoxide’s substituent. The structure of the complex also shows that the combination of the C 2-symmetry of the catalyst with the meso-symmetry of the substrate is ideal as anticipated in the initial catalyst design. For me, the complex was also of exceptional predictive value. The substrate binding in 3·4 very strongly suggested that ‘pseudo meso-epoxides’, i.e., cis-1,2-disubstituted epoxides with ‘similar’ substituents should be opened with high selectivity. It remained to be explored how far ‘similarity’ could be pushed before selectivity breaks down.

By breaking the symmetry, the substrates become chiral. Therefore, the complexation and reactivity of the complexes are more intricate than for meso-epoxides in a number of respects. With chiral substrates, the substrate–catalyst complexes of the respective enantiomers of the epoxide with the enantiomerically pure titanocene are diastereoisomers and will, therefore, show different reactivity.

If one of the enantiomers is opened much faster than the other, a kinetic resolution will result. This is not what we wanted. What we hoped to find was a reaction with both enantiomers of the substrate being ring-opened with a similar rate and with the same mechanism of steric differentiation as with 3·4a. This scenario should lead to the highly selective transformation of one enantiomer of the epoxide substrate into one regioisomer of the product and to the transformation of the other enantiomer into the other regioisomer of the product with a similar and high selectivity. For racemic substrates this will result in the formation of two highly enantioenriched products in a ratio close to 50:50. We named these epoxide openings regiodivergent. Simple desymmetrizations of meso-epoxides constitute special cases of regiodivergent reactions, where the regioisomeric products are enantiomers.

Fan Chun-An, now a very successful professor in Lanzhou in China, and Florian Keller decided to take the challenge and develop such highly selective regiodivergent epoxide openings.[17] They soon realized that breaking symmetry comes at a cost in practice. With racemic substrates, not only two but four products can be obtained (both enantiomers of the two regioisomeric ring-opening products). Performing the reactions with enantiomerically pure substrates is less laborious. Only one substrate is used.

Zoom Image
Scheme 2 Reagent-controlled regiodivergent epoxide opening with 3 as the catalyst

As shown in Scheme [2], Fan Chun-An and Florian Keller were successful in demonstrating that with ‘unbiased’ enantiomerically enriched epoxides, the regioselectivity of the ring opening was essentially controlled by the absolute configuration of the catalyst with a low influence of substrate control on the regioselectivity of ring opening. Of course, the reaction also worked with racemic substrates. These cases constitute examples of parallel resolution. There is a more provocative (or even frivolous) experiment: the reaction of an enantiomerically pure substrate with racemic 3. We did not do it with the simple substrates, but the idea is less crazy than it may seem and is useful in radical arylation. This will be discussed in Section 4.

Our first examples constituted ‘only simple’ reductions of ‘simple epoxides’. We believed that an important argument in favor of our concept is its application in reactions that increase molecular complexity. With radicals this is easily achieved by trapping them through C–C bond formation. The 5-exo cyclization is the most common reaction in this respect.

Zoom Image
Scheme 3 Combination of the REO with 5-exo cyclizations

Matthias Otte and Shi Lei (Scheme [3]) combined regiodivergent epoxide opening with 5-exo cyclization for their synthesis of exo-methylenecyclopentanes from suitably functionalized ‘Sharpless’ epoxides.[18] Compared to the epoxide substrates, the cyclopentane products are structurally more complex than the substrates. Moreover, they are valuable starting materials for further functionalization and are of interest for natural product synthesis and for the preparation of antiviral compounds. I am somewhat ashamed to admit that we did not discuss or even mention that the ‘other’ regioisomers, that were only a 1,2-diols, were not really interesting at all. So, honestly speaking, the goal of increasing molecular complexity was only achieved for ‘half of the REO’.

An analysis of the ratios of products obtained and their enantiomeric purity revealed one of the limits of ‘similarity’ with derivatives of Sharpless epoxides. The formation of the ‘1,2-diol’ product is preferred over the formation of the ‘1,3-diol’ product. This is due to the electronic destabilization of radical A by the two substituents bearing a C–O bond, and is not present in radical B. This influence operates in addition to the high steric differentiation of the substrate complexes with 3 and leads to the favored formation of B, as further verified by Peter Karbaum and Dave Schmauch (Scheme [4]).[19] The experimental results are fully in agreement with a computational study that was performed by the group of Frank Neese who were working in Bonn at that time. Peter Karbaum also made the important discovery that the electronic effects were negligible for epoxides derived from homoallyl alcohols. Therefore, it seemed possible to obtain 1,3- or 1,4-diols from these substrates with high selectivity.

Zoom Image
Scheme 4 Low selectivity of the REO of ‘Sharpless’ epoxides catalyzed by 3

At this time, Nico Funken took over the project and Felix Mühlhaus joined as an M.Sc. student. One of the first starting materials was derived from (R)-ricinolate and was easily accessible on multigram scale. Depending on the absolute configuration of the catalyst, either the 1,3-diol or the 1,4-diol could indeed be prepared with high regioselectivity as essentially enantiomerically pure compounds (Scheme [5]).[20] In addition, Nico Funken increased the activity of the catalytic system substantially by cleverly exploiting the use of bromide and chloride ligands in 3. Somewhat unfortunately, we had to use the unpopular but chemically excellent Bu3SnH as the hydrogen atom donor.

Zoom Image
Scheme 5 Regiodivergent epoxide opening for the synthesis 1,3- or 1,4-diols from the same substrate. Lu·HCl = 2,6-lutidine hydrochloride.

A very important point is that in general the substrates can be synthesized enantiomerically pure in just three steps [alkyne addition to enantiomerically pure epoxides to yield a homopropargylic alcohol, Lindlar hydrogenation to the (Z)-homoallylic alcohol, and diastereoselective, VO(acac)2-catalyzed epoxidation] in a modular approach.

Of course, our method can be used for target-oriented synthesis of specific molecules. This classical approach does not make full use its potential, though, because only half of the accessible products are of interest. The essence of the REO is that we can prepare 1,3-diols and 1,4-diols depending on the choice of the absolute configuration of the catalyst. Such a method is of high interest for applications in diversity-oriented synthesis (DOS)[21] that aims toward generating libraries of structurally and functionally diverse compounds. Providing branching points for DOS is especially attractive and the REO can do just that.

Did we finally arrive at an interesting and important reaction? Not quite. The REO described above is not very sustainable, it requires Bu3SnH, and one can argue that the molecular complexity of the 1,3- and 1,4-diols is not significantly higher than those of the β-hydroxy epoxides. These issues were addressed in our REO arylation that introduces atom economical C–C bond formation to the field of REO.

Before discussing this reaction, it is necessary to have a closer look at the aspects of sustainability of titanocene catalysis and at epoxide arylation.


# 3

Atom Economy and Epoxide Arylation

One of the critical issues of titanocene catalysis is the sustainability of most applications. In the reaction shown above, we did not bother too much. Obtaining high selectivities was the major goal. However, we also developed interest in sustainable and catalytic radical chemistry. The CRC 813 ‘Chemistry at Spin Centers’ (established in 2009) provided an ideal framework for such endeavors. The general idea that we called ‘catalysis in single electron steps’ was conceptually ambitious because we wanted to merge the advantages of radical chemistry, such as mild reaction conditions and the exceptional potential for forming C–C and C–X bonds, with those of transition-metal catalysis, especially the control of stereochemical aspects of radical generation and trapping.

Zoom Image
Scheme 6 Catalysis in single electron steps for the synthesis of tetrahydrofurans

Our first example highlights a sequence featuring the general idea (Scheme [6]).[22] The catalytic cycle can be seen in two ways. The obvious way is to look at it as from the point of view of radical chemistry. Epoxide opening is an electron-transfer reaction from the metal to the substrate that is followed by a 5-exo cyclization. The tetrahydrofuran-forming reaction is a little less intuitive but, in the end, it is just an organometallic oxygen rebound.[23] It is beyond the scope of this account to even attempt a discussion of oxygen rebounds. However, the ability of nature to do it (the P450 enzymes!) was very helpful in persuading critics (and reviewers) about the validity of my analysis. One should not forget that the final step of many catalytic (or stoichiometric) epoxidation reactions with metaloxo species can (and probably should) also be considered as an oxygen rebound.

The second view that ‘looks at things from the perspective of titanocene’ seemed much more inspiring to me. Here, epoxide opening constitutes an oxidative addition of Cp2Ti(III)Cl to the epoxide. The 5-exo cyclization can be seen as a ‘radical translocation’ setting the stage for the final reductive elimination of Cp2Ti(III)Cl, the active catalyst, that generates the tetrahydrofuran. Thus, it is simply ‘normal catalysis’ with the metal shuttling between neighboring oxidation states in single electron steps. This very simple analysis directly leads to the desired use of radicals as the key intermediates. The oxidative addition generates the open-shell organic radical and the reductive elimination transforms it into the desired closed-shell organic product. Reaction design is simple from here on: Find a good radical translocation that allows a reductive elimination and you have a novel cycle! Finally, two points that our M.Sc. students like. First, the reaction is atom-economical and therefore ‘simply’ an isomerization. Second, the driving force for the isomerization is ‘visible’. It is the release of epoxide ring strain.

In 2008 the Wipf group reported a very interesting and potentially useful titanocene-catalyzed radical arylation that features an aromatization of the radical σ-complex via its oxidation with O2 that makes the use of stoichiometric amounts of Coll·HCl and Zn necessary.[24] I was not convinced by the correctness of the mechanism for two reasons. First, we had tried to incorporate O2 in titanocene-catalyzed reactions and had always failed quite miserably. Second, armed with the ideas described above, the arylation is an isomerization and so the only thing we need for atom-economical catalysis is a suitable reductive elimination.

As Maike Behlendorf, Daniel von Laufenberg, and André Fleckhaus found (Scheme [7]), in the presence of 10 mol% of Cp2TiCl a 96% yield of the desired product was obtained and no Coll·HCl was needed. It was ‘normal catalysis’, after all.[25] A reduction of the catalyst loading to 1 mol% was possible after the addition of 5 mol% of Coll·HCl. The additive is essential for stabilizing Cp2Ti(III)Cl by forming [Cp2Ti(III)Cl2]– as the resting state of the catalyst, as shown by Christian Kube in Bonn and Dhandapani V. Sadasivam at Lehigh University. How is the catalytic cycle closed if no O2 is needed? What we need to achieve is a reduction of Ti(IV) to Ti(III) [by electron transfer (ET)] and a protic cleavage of the Ti–O bond (a protonation). Combining these two steps results in a proton-coupled ET (PCET), an unconventional reaction for homolytic bond activation.[26]

Zoom Image
Scheme 7 Catalysis in single electron steps for the radical arylation of epoxides

The mechanistic investigations that combined kinetics (Dhandapani V. Sadasivam and Godfred D. Fianu, Bob ­Flowers’ group), synthetic studies (Tobias Dahmen, Antonius Michelmann and Meriam Seddiqzai, my group) and calculations (Rebecca Sure, Stefan Grimme’s group) showed that electron-deficient titanocenes lead to faster reactions and that radical addition is slightly slower than a 5-exo cyclization.[27] This implies that epoxide opening and radical addition are not the rate-determining steps, but rather reductive elimination via PCET. As found by Ruben Richrath and Theresa Liedtke in Bonn, and Daniel G. Enny at Lehigh, Cp2Ti(OMs)2 and Cp2Ti(OTs)2 are easy to prepare and both are more stable and active catalysts than Cp2TiCl2-derived complexes.[28]

The reaction is broad in scope concerning the synthesis of pharmaceutically important indolines and tetrahydroquinolines but, as Sven Hildebrandt has shown, can also be used for the preparation of the equally attractive dihydropyrrolizines and tetrahydroindolizines.[29]


# 4

REO Arylation

At this stage, we had understood how to carry out the radical arylation under robust and sustainable conditions and had a good idea about the substrate scope of the reaction. It was time to move on to regiodivergent catalysis. Felix Mühlhaus and Hendrik Weißbarth decided to merge the radical arylation with regiodivergent epoxide opening (REO) by making either tetrahydroquinolines or indolines from identical substrates (Scheme [8]).[1]

Zoom Image
Scheme 8 Concept of the REO arylation

Three points are critical for an efficient process. First, epoxide opening should occur with high catalyst-controlled regioselectivity. Second, the diastereoselectivity of radical addition to the arene should be high. Third, in order to be useful for synthetic applications, the preparation of the substrates must be easy to carry out in a modular fashion.

Zoom Image
Scheme 9 Synthesis of the substrates for the REO arylation. HKR = hydrolytic kinetic resolution.

The general solution to the third issue was found by ­Tobias Dahmen (Scheme [9]) and starts from readily available (E)-allylic alcohols. After bromination and treatment with a base, a bromo epoxide was obtained that could be readily resolved by Jacobsen’s excellent kinetic resolution.[5b] Epoxide opening with an amine (or amide) nucleophile and subsequent substitution of bromine (both via an SN2 mechanism) provided the desired starting materials. The bromo epoxide and the final product can be obtained in one-pot reactions. Thus, our sequence requires only three steps from an allylic alcohol to the enantiomerically pure substrate.

Felix Mühlhaus investigated the REO arylation with 3 (Scheme [10]). We were somewhat worried that, similar to the REO of Sharpless epoxides (Scheme [3]), there would be a strong influence of the CH2OR substituent of the epoxide on the regioselectivity of ring opening. With Cp2TiCl2 this was indeed the case. However, with Kagan’s catalysts 3 and ent-3,[10] the electronic influence of the NPh2 group on the regio­selectivity was negligible (93:7 with 3 and 10:90 with ent-3). To our pleasant surprise, Felix and Hendrik found that the relatively low diastereoselectivity in tetrahydroquinoline formation could be increased by replacing the Cl with an OTs ligand in the catalyst. We have proposed transition-state models explaining this effect.

Zoom Image
Scheme 10 Examples demonstrating the scope of the REO arylation e.r. = enantiomeric ratio

We then proceeded to examine the scope of tetrahydroquinoline (THQ) and indoline synthesis. It turned out that N-alkyl substitution leads to much better results in THQ synthesis, especially with respect to diastereoselectivity. The alkyl group may be CH3, primary, secondary, and tertiary. Cyclopropyl substituents are also tolerated. For indolines, however, NAr2 substitution is favorable for the regio­selectivity of ring opening and the diastereoselectivity of radical addition. The indolines can be obtained as single enantiomers and diastereoisomers after straightforward purification by silica gel chromatography. The reaction tolerates a large variety of functional groups as substituents on the arene that are attractive for pharmaceutical applications, such as carbonyl, alkyl, aryl, F, Cl, Br, and even I.

If one is interested in a particular indoline or THQ target, our method should be both interesting and useful as the syntheses are short and tolerate many functional groups. I believe that applications in such target-oriented synthesis (TOS) are not the most attractive of our regiodivergent arylation. When we designed the regiodivergent epoxide arylation, we were still ‘trapped’ in a TOS-biased thinking. We wanted to make one of two possible compounds from a single substrate with high catalyst control. However, our reaction is more interesting as a branching point for diversity-oriented synthesis (DOS)[21] because we can make two compounds from one substrate with high selectivity. This raises the question of what the easiest way to do so is. The obvious answer is to do two reactions and then two purifications to access both the indoline and the THQ. Doing both reactions in one flask, i.e., reacting the enantiomerically pure substrate with the racemic catalyst, gives the same products and does so in the same ratio. However, only one reaction and only one purification is necessary. From a DOS-perspective, using the titanocene catalyst in racemic form, provided that no scrambling of the ligands occurs, is therefore not frivolous (or crazy) at all but is to be preferred over the use of enantiomerically pure catalysts. Do it, it works! I leave the assessment of the reaction with a racemic substrate and a racemic catalyst to you.


# 5

Words about the Wording

The following explanatory prose is aimed at providing more general explanations and classifications of our concepts by comparing them with related reactions from other groups and with common vocabulary on stereochemistry. It therefore puts the REO arylation in a broader perspective.

I subsume the catalytic cycles of tetrahydrofuran synthesis and the REO arylation under the term of ‘catalysis in single electron steps’.[30] This seems reasonable because the cycles strongly resemble those of Pd-catalyzed cross-couplings or Rh-catalyzed hydrogenations.[31] However, Ti shuttles[21c] between two oxidation steps in single electron steps and not in two electron steps as Rh and Pd do. The analogy becomes even more obvious when one takes into account that Hegedus and Söderberg have pointed out that ‘the terms ‘oxidative addition’ and ‘reductive elimination’ are generic, describing an overall transformation, but not the specific mechanism by which the transformation occurs’.[32] Accordingly, the first step of the cycle (radical generation from the epoxide and Cp2Ti(III)Cl) is an oxidative addition. To highlight the shuttling of titanium between Ti(III) and Ti(IV), we call such steps oxidative addition in single electron steps. It comes as no surprise that the final step of the cycle (transformation of the radical into the closed-shell organic product with concomitant reduction of Ti and release of Cp2Ti(III)Cl) is a reductive elimination. In the radical translocation step, any of the many great radical reactions can be inserted for selective bond formation. Is there anything special about looking at catalytic radical chemistry as described above? In my opinion, there really isn’t. If you have a metal that easily shuttles between oxidation states in single electron steps, radicals are your intermediates of choice for efficient catalysis.

Are the titanocenes in the context of epoxide opening unique in enabling this type of catalysis or is the concept too special? The answer is ‘no’, of course. Song Lin at Cornell has recently reported titanium-catalyzed ‘radical relay catalysis’ that transforms cyclopropyl ketones or N-acyl aziridines and olefins into polysubstituted cyclopentanes and N-acyl pyrrolidines, respectively.[33]

Bas de Bruin in Amsterdam and Peter Zhang have made use of Co(II) complexes of porphyrins in metalloradical catalysis (MRC).[34]

Zoom Image
Scheme 11 Metalloradical catalysis (MRC) for the cyclopropanation of acrylates. por = porphyrin.

The key to the success of their MRC is that that the reactions of these complexes with diazo compounds should be viewed in terms of Co(III)-carbene radicals rather than as Co(II) carbenes (Scheme [11]).[34] In cyclopropanation reactions, these metal-bound radicals can be used in radical translocation steps such as additions to acrylates. Ring closure is occurring via a homolytic substitution at the Co(III)–C that leads to the release of the Co(II)–porphyrin complex. One can apply the nomenclature described for Ti. Co(II) oxidatively adds to the substrate, the radical is translocated, and, finally, a reductive elimination of Co(II) releases the organic product, the cyclopropane. Metalloradical catalysis (MRC) is an equally valid description, of course, that can also be applied to Ti(III) catalysis. The Co(II) chemistry with diazo compounds has been applied in many excellent reactions. Moreover, azides are also very interesting and useful substrates. For both substrates enantioselective applications have been developed. This highlights the power of merging the advantages of organometallic chemistry with radical chemistry once again.

David Procter at Manchester has shown how SmI2 can be used as an electron-transfer catalyst in reactions featuring ‘cyclization cascades by radical relay’.[35]

Zoom Image
Scheme 12 SmI2-catalyzed cyclization cascades by radical relay

In these reactions (Scheme [12]), the oxidative addition is simply called electron transfer and the reductive elimination back-electron transfer. This labelling emphasized the role of Sm in the cycle and highlights its spectacular aspect, the reduction of Sm(III) by an organic ketyl radical. Since the general perception of Sm reagents in radical chemistry is their role as reductants, simply regarding the back-electron transfer to Sm(III) as a reductive elimination is probably a little too dispassionate or outright boring!

Finally, a comment on the implications of using the term regiodivergent. Nico Funken, Yong-Qiang Zhang and I have explained for which reactions we think the term regiodivergent is appropriate.[36] This implies that ‘our’ regiodivergent reactions are the more general case of desymmetrization reactions. In the desymmetrization reaction of meso-epoxides, a successful catalyst selectively activates and eventually cleaves one of the two enantiotopic C–O bonds. The nomenclature is consistent. Cleavage of enantiotopic bonds leads to enantiomeric products that are also the regioisomers of ring opening. How should the C–O bonds in the epoxides in Schemes 2, 3, 4, 5, and 10 be classified? The substrates have no element of symmetry (other than a C 1 axis). By the standard definitions, the C–O bonds are heterotopic. However, after having such a neat selectivity for cleaving these bonds (after all this time), calling them heterotopic is simply too cynical.

One could call the two C–O bonds pseudo-enantiotopic, because the substrates are ‘almost’ meso-compounds. However, this misses the point that desymmetrizations are special cases of our regiodivergent catalysis and constitute a clear case of the tail wagging the dog. We have the more general case of high catalyst control in the regioselectivity of the cleavage of the C–O bonds of the epoxide. Therefore, we call the C–O bonds in our substrates regiotopic since the products of our reactions are regioisomers.


# 6

Conclusion

It has been quite a long way (in terms of time, life-time, and chemistry) from the conclusion of our first Account that ‘asymmetric versions of the reaction are also being investigated in our labs in Göttingen and first results are very promising, indeed’ to the REO arylation. We have not only extended the scope of the catalyst-controlled regioselective opening of cis-1,2-disubstituted epoxides well beyond meso-epoxides, but we have also understood how to make more complex products from these reactions. En route, we have understood how to make Ti shuttle between the oxidation states III and IV in ‘catalysis in single electron steps’ and how to make the reactions attractive, not only for applications in target-oriented synthesis (TOS), but also as branching points in diversity-oriented synthesis (DOS). The potential of these developments is substantial. Therefore, things are probably not only interesting but are important!

Finally, I dare to repeat that ‘first examples are very promising indeed’,[37] but this time in more areas of research than almost 20 years ago. A cool conclusion after all this time with epoxides and titanocenes!


#
#

Acknowledgment

The generous financial support from the DGF, the Alexander von Humboldt-Stiftung, the Jürgen Manchot Stiftung, the Studienstiftung des Deutschen Volkes, the Konrad-Adenauer-Stiftung, the Hans-Böckler-Stiftung, and the Evangelisches Studienwerk Villigst is gratefully acknowledged. Money is important but not everything. I have thoroughly enjoyed being a member of the Gansäuer group for more than 20 years. It was great fun inside and outside of the lab! However, the most important people to me are my wife and my two daughters. They show me every day that life has much more to offer than being a ‘science-nerd’. Thanks Monika, Sophia, and Anna Lena!



   Permissions and Reprints All articles of this category
Zoom Image
Zoom Image
Scheme 1 Structures of the catalysts investigated in the desymmetrization of meso-epoxides
Zoom Image
Figure 1 X-ray structures of 2 and 3 [12]
Zoom Image
Figure 2 Structure (left) and schematic representation (right) of the catalyst–substrate complex 3·4
Zoom Image
Scheme 2 Reagent-controlled regiodivergent epoxide opening with 3 as the catalyst
Zoom Image
Scheme 3 Combination of the REO with 5-exo cyclizations
Zoom Image
Scheme 4 Low selectivity of the REO of ‘Sharpless’ epoxides catalyzed by 3
Zoom Image
Scheme 5 Regiodivergent epoxide opening for the synthesis 1,3- or 1,4-diols from the same substrate. Lu·HCl = 2,6-lutidine hydrochloride.
Zoom Image
Scheme 6 Catalysis in single electron steps for the synthesis of tetrahydrofurans
Zoom Image
Scheme 7 Catalysis in single electron steps for the radical arylation of epoxides
Zoom Image
Scheme 8 Concept of the REO arylation
Zoom Image
Scheme 9 Synthesis of the substrates for the REO arylation. HKR = hydrolytic kinetic resolution.
Zoom Image
Scheme 10 Examples demonstrating the scope of the REO arylation e.r. = enantiomeric ratio
Zoom Image
Scheme 11 Metalloradical catalysis (MRC) for the cyclopropanation of acrylates. por = porphyrin.
Zoom Image
Scheme 12 SmI2-catalyzed cyclization cascades by radical relay