Enantioselective Intermolecular C-H Amination Directed by a Chiral Cation.

The enantioselective amination of C(sp3)-H bonds is a powerful synthetic transformation yet highly challenging to achieve in an intermolecular sense. We have developed a family of anionic variants of the best-in-class catalyst for Rh-catalyzed C-H amination, Rh2(esp)2, with which we have associated chiral cations derived from quaternized cinchona alkaloids. These ion-paired catalysts enable high levels of enantioselectivity to be achieved in the benzylic C-H amination of substrates bearing pendant hydroxyl groups. Additionally, the quinoline of the chiral cation appears to engage in axial ligation to the rhodium complex, providing improved yields of product versus Rh2(esp)2 and highlighting the dual role that the cation is playing. These results underline the potential of using chiral cations to control enantioselectivity in challenging transition-metal-catalyzed transformations.

The ability to form new C–N bonds in a direct and efficient manner is crucially important due to their ubiquity in organic molecules. Traditional disconnections are increasingly supplemented by methods which can use far less reactive C–H bonds, enabling powerful alternative retrosynthetic strategies.[1] One of the most widely used and versatile involves the insertion of catalytically generated rhodium nitrenoids into C(sp3)–H bonds.[2] The original catalysts used for this purpose were dirhodium tetracarboxylates (also referred to as paddlewheel complexes),[3] and extensive development of this methodology has been undertaken by Du Bois and co-workers in particular.[2a] This has culminated in the development of the versatile and robust strapped dicarboxylate catalyst Rh2(esp)2 which can perform rhodium-catalyzed C–H amination intermolecularly on benzylic, tertiary, and, in some cases, secondary alkyl C–H bonds (Figure a).[2a,4] In many instances, C–H amination leads to the introduction of a new stereocenter and efforts to render Rh(II)-catalyzed C–H aminations enantioselective have been ongoing since its early development.[5] Due to the ready availability of chiral carboxylic acids, their incorporation into paddlewheel complexes has constituted the main strategy, encompassing important contributions from Hashimoto,[6] Müller,[7] Davies,[8] and Dauban,[9] among others (Figure b, left). Additionally, Du Bois developed a chiral carboxamidate variant for enantioselective intramolecular amination (Figure b, middle).[10] Despite these advances, as well as related ones employing alternative transition metals[11] and enzymes,[12] intermolecular C–H amination via nitrene transfer still remains extremely challenging to achieve asymmetrically. For rhodium dimers bearing chiral carboxylate ligands, the chiral information is located at a considerable distance from the reactive axial site. Although very successful for enantioselective carbene C–H insertions,[13] for nitrene insertions the development of fundamentally different strategies is clearly warranted. Given that Rh2(esp)2 is the current state-of-the-art catalyst for nonenantioselective intermolecular C–H amination, the development of a chiral variant could be transformational but there are few structural opportunities to achieve this.[4b,14] In a creative strategy, Bach and co-workers tethered the bridging aryl ring of (esp), through an alkyne linker, to a chiral lactam (Figure b, right).[15] A dual hydrogen bonding interaction with the substrate permitted benzylic C–H amination with up to 74% ee.

Figure 1

Background to enantioselective C–H amination using Rh dimers (a, b) and this work (c, d).

We recently outlined a strategy for inducing asymmetry in reactions that use ligand scaffolds which are particularly challenging to render chiral in the conventional manner. In our approach, the ligand is made anionic through the attachment of a sulfonate group which in turn allows association of a chiral cation with which to exert enantiocontrol (Figure c).[16] This strategy provides an opportunity to unite privileged chiral cations with the diverse reactivity of transition metal complexes.[17] In our first study a distally sulfonated bipyridine ligand was associated with a quinine-derived cation to impart enantiocontrol in iridium-catalyzed arene borylation.[18] Quaternized cinchona alkaloids provide a well-defined chiral pocket with ample opportunity for attractive noncovalent interactions between the substrate and the rich functionality of the cation.[19] Seeking to apply this strategy to C–H amination we first synthesized an (esp) analogue bearing a methylenesulfonate group on each bridging benzene ring. These anionic handles would then be used to associate with chiral cations to form a “sulfonesp” family of ion-paired catalysts (Figure d).

The “sulfonesp” scaffolds were readily synthesized in a three-step sequence comprising dialkylation using ester enolates, displacement of the remaining benzyl bromide with sodium sulfite, and ester hydrolysis to the corresponding diacid (Figure a). After assembly of the rhodium dimers, the bisligated complexes were isolated as the bistetrabutylammonium salts and it proved straightforward to introduce chiral cations via intermediate protonation using Amberlite IRC120 H. This accessed a series of “sulfonesp” scaffolds with varied geminal dialkyl substitution, both acyclic (A) and cyclic (B–E) (Figure b), a steric parameter on the ligand that we anticipated could be used to tune enantioselectivity. Notably, chirality is introduced in the final ion exchange, enabling rapid access to libraries of ion-paired catalysts.[20] We initially synthesized the gem-dimethyl ligand scaffold (A) in combination with dihydroquinine-derived (DHQ-derived, 1) and dihydroquinidine-derived (DHQD-derived, 2a) cations that bore the specific bulky quaternizing benzyl group that had been optimal in our previous work (Figure c).[18] Intriguingly, we observed a striking solution color change from green to red once the chiral cations were incorporated. This strongly suggested axial ligation of rhodium, with the quinoline nitrogen of the cations constituting the most likely ligand. UV–visible studies lent strong support to this hypothesis: comparing the λmax of Rh2(D)2·(2a)2 (536 nm) with Rh2(esp)2 (655 nm) in 1,3-difluorobenzene suggests a significant difference in their respective HOMO–LUMO energy gaps (see Supporting Information (SI) for further details).[21] While such binding could prevent nitrenoid formation if the binding in solution were too strong, we anticipated that a weaker, reversible interaction could actually be beneficial, potentially protecting the rhodium dimer from decomposition pathways and extending the catalyst lifetime, as has been shown in a number of recent studies.[4d,22]

Figure 2

Synthesis (a) and systematic variation of ion-paired “sulfonesp” ligands on both ligand scaffold (b) and cation (c).

We began our investigations using 4-phenybutan-1-ol (6a) as a challenging test substrate (Table ). This contains prochiral benzylic C–H bonds and a hydroxyl functionality that could feasibly hydrogen bond with the catalyst sulfonate group to provide organization at the transition state. The direct rhodium-catalyzed intermolecular amination of substrates containing unhindered primary alcohols has little precedent, yet the resulting chiral aminoalcohol derivatives could be of great synthetic utility, particularly since they are an oxidation away from γ-aminobutyric acids.[23] Additionally, the transformation would give rise to products not currently accessible using Du Bois’ enantioselective intramolecular amination methodology involving cyclization of sulfamate esters en route to 1,3-amino alcohols.[10] Although Rh2(esp)2 gave a very low yield (9%) on this substrate under the evaluation reaction conditions, we were pleased to observe that this increased significantly (43%) when Rh2(A)2·(1)2 was used. Further, a low but encouraging ee of 33% was measured (Table , entries 1 and 2). Remarkably, when using complex Rh2(A)2·(2a)2 containing the pseudoenantiomeric DHQD-derived cation, the ee increased drastically from 33% to −71% (entry 3). Such divergence in the enantiomeric excesses afforded by the pseudoenantiomers is intriguing but has been noted in other systems.[24] This prompted us to evaluate another set of diastereomers of the cinchona alkaloid family, namely the epi-DHQ-derived (3) and epi-DHQD-derived (4) cations, in which the hydroxyl-bearing stereocenter is inverted on each. In these cases, the ee outcomes were poor (entries 4 and 5) so we continued optimization with the DHQD-derived cations. We were pleased to discover that switching the oxidant from PhI(OPiv)2 to iodosobenzene (PhIO) increased both conversion and ee (Table , entry 6). Despite the moderate yield, full conversion of starting material was observed along with a number of uncharacterized byproducts. A switch to the lower melting 1,3-difluorobenzene solvent enabled us to reduce the temperature to −25 °C, which in turn allowed for a more controlled reaction to give the product in an excellent 83% yield and −81% ee (entry 7). We next evaluated dimer scaffolds B–E to systematically explore steric changes near to the active site which we anticipated might lead to subtle variations in cation and substrate positioning in the enantiodetermining transition state (entries 8–11). This revealed that the cycloheptyl “sulfonesp” scaffold D provided both optimal yield (90%) and ee (−90%) in the complex Rh2(D)2·(2a)2. Finally, we returned to evaluate a selection of other quaternizing groups on the DHQD framework in conjunction with optimal scaffold D. Replacing the t-Bu groups at the periphery of the teraryl unit with CF3 (2b) or removing them completely (2c) was detrimental to the ee (entries 12 and 13), as was removing the outer two aryl rings of the teraryl unit (entries 14 and 15). Under the optimal conditions, the ion-paired catalysts greatly outperformed Rh2(esp)2, which delivered only a 17% yield (entry 16), underlining the importance of the chiral cation in improving reaction yield in addition to its pivotal role in enantioinduction. Indeed, when 4-phenylbutan-1-ol was subjected to the current state-of-the-art conditions for intermolecular amination using either PhOSO2NH2 or DfsNH2 as the aminating agents and Rh2(esp)2 as the catalyst, the desired product was not observed.[4d] Only by use of the more reactive perfluorinated sulfamate ester 5 could ∼20% crude 1H NMR yield be obtained, emphasizing that 6a is a challenging substrate for C–H amination (see SI).[25]

Table 1

Reaction Optimizationa

Entry	Catalyst	Oxidant	Temp (°C)	Solvent	Yield (%)	eeb (%)
1	Rh2(esp)2	PhI(OPiv)2	–10	1,4-DFB	9	Rac.
2	Rh2(A)2·(1)2	PhI(OPiv)2	–10	1,4-DFB	43	33
3	Rh2(A)2·(2a)2	PhI(OPiv)2	–10	1,4-DFB	35	–71
4	Rh2(A)2·(3)2	PhI(OPiv)2	–10	1,4-DFB	13	Rac.
5	Rh2(A)2·(4)2	PhI(OPiv)2	–10	1,4-DFB	21	+26
6	Rh2(A)2·(2a)2	PhIO	–10	1,4-DFB	46	–87
7	Rh2(A)2·(2a)2	PhIO	–25	1,3-DFB	83	–81
8	Rh2(B)2·(2a)2	PhIO	–25	1,3-DFB	60	–81
9	Rh2(C)2·(2a)2	PhIO	–25	1,3-DFB	75	–87
10	Rh2(D)2·(2a)2	PhIO	–25	1,3-DFB	90c	–90c
11	Rh2(E)2·(2a)2	PhIO	–25	1,3-DFB	71	–86
12	Rh2(D)2·(2b)2	PhIO	–25	1,3-DFB	46	–53
13	Rh2(D)2·(2c)2	PhIO	–25	1,3-DFB	58	–76
14	Rh2(D)2·(2d)2	PhIO	–25	1,3-DFB	57	–82
15	Rh2(D)2·(2e)2	PhIO	–25	1,3-DFB	58	–74
16	Rh2(esp)2	PhIO	–25	1,3-DFB	17c	Rac.c

Reactions performed on 0.1 mmol scale with respect to 6a using 1.2 equivalents of 5. Reaction concentration = 0.2 M. Yields determined by 1H NMR with reference to internal standard.

ee determined by chiral SFC analysis of the crude reaction.

Data corresponds to the isolated sample. DFB = difluorobenzene.

With the optimized conditions in hand we evaluated the tolerance to various arene substituents (Scheme ). An ester at the meta position was well tolerated (7b) as were methyl groups at the ortho and meta positions (7c, 7d). Substrates with fluorine atoms in all three positions also gave very high levels of enantioselectivity (7e–7g). With meta-substituted substrates (7h–7l), we were pleased to see that high enantioselectivity was maintained although in these cases, increasing the catalyst loading to 3.0 mol % was beneficial for conversion. Substitution at the ortho position with chlorine gave a reduced yield (7m) although the enantioselectivity remained high. Substitution at the para position gave a slightly reduced enantioselectivity in the case of methoxy (7n), but not chloride (7o). Finally, a range of disubstituted arenes (7p–7v) were compatible as well as naphthalene (7w) and thiophene (7x) containing substrates. For products 7q and 7r, a higher temperature was used for improved conversion and the solvent was switched from 1,3-difluorobenzene to 1,4-difluorobenzene since the latter had given a slightly improved enantioselectivity in initial reaction optimization (Table , entries 6 and 7).

Scheme 1

Reaction Scope Exploration

Reaction performed with 3.0 mol % Rh2(D)2·(2a)2.

Reaction performed at −10 °C using 1,4-DFB in place of 1,3-DFB.

Reaction product 7a was readily transformed into protected 2-arylpyrrolidine 8 using Mitsunobu chemistry (Scheme a). N-Deprotection of 8 allowed assignment of the absolute stereochemistry of the products by comparison of the optical rotation of 9 with literature values (all other amination products were assigned by analogy). Our earlier observation that the precise diastereomer of the cinchona alkaloid scaffold used greatly impacted the enantioselectivity was curious, but also a practical limitation if the opposite product enantiomer is required. Assuming that the ethyl group in DHQ-derived 1 causes an unfavorable steric interaction at the transition state, we removed it by devinylation of quinine.[24] We were pleased to find that the resulting catalyst Rh2(D)2·(10)2 gave the product ent-7a with almost exactly the opposite sense of enantioinduction and with only a small reduction in yield (Scheme b). To probe the importance of the proposed hydrogen bonding between the substrate hydroxyl and the catalyst sulfonate, we evaluated the amination of phenylbutane (Scheme c, left). This showed drastically reduced reactivity and enantioselectivity suggesting that the attractive interaction is crucial for both outcomes. We also carried out the amination using Rh2(esp)2 in combination with 2a·Br to examine the effect of severing the ionic link between ligand and cation. This resulted in poor enantioselectivity (19% ee, Scheme c, right). Interestingly, the yield was significantly improved compared with Rh2(esp)2 alone (51% vs 17%) which provides support for beneficial axial ligation by the quinoline of the cation, even when the cation is not associated with the ligand. Further support for this fortuitous benefit provided by the cinchona alkaloid-based chiral cations was provided by the poor yield (7%) obtained using an ion-paired ligand bearing tetrabutylammonium in place of the quinoline-containing chiral cation.

Scheme 2

Practical Considerations and Control Experiments

We next tested substrate 12, in which the hydroxyl is replaced by a carboxylic acid to evaluate its ability to interact productively with the catalyst. Here PhIO as the oxidant led to only racemic products, but switching to PhI(OPiv)2 afforded the product with an encouraging level of enantioenrichment (78% ee), albeit in low yield in these initial investigations (Scheme a). We also evaluated shorter (three carbon) and longer (five carbon) chain alcohols against our five “sulfonesp” scaffolds A–E, all using cation 2a. This revealed that, for phenylpropanol, the same scaffold (D) that was optimal for phenylbutanol was best and cyclobutane-containing B was poorest (Scheme b, 14). However, for the longer chain phenylpentanol, scaffold B was superior to all others (Scheme b, 15). While still preliminary, these encouraging results provide a compelling demonstration that the modularity of our ion-paired “sulfonesp” ligands, in terms of both ligand scaffold and cation, will facilitate matching them with future substrates of interest. We also tested other functional groups in place of hydroxyl, which gave poor outcomes (see SI for details).

Scheme 3

Further Substrate Exploration

In conclusion, we have developed a family of ion-paired chiral catalysts for rhodium-catalyzed C–H amination based on the (esp) ligand scaffold and have applied them successfully to the enantioselective intermolecular C–H amination of 4-arylbutanols. Furthermore, the optimal ion-paired catalyst also results in significantly improved yields compared with Rh2(esp)2. We believe that this is due to a combination of axial coordination by the chiral cation and a network of noncovalent interactions between ligand and substrate which promote the desired benzylic amination. These results form the basis of a catalyst design principle that we anticipate, with further development, should be applicable to intermolecular amination reactions of other challenging substrate classes. More broadly, this demonstrates the potential of using ion-paired ligands bearing chiral cations to tackle challenging transition-metal-catalyzed reactions.