‘Goldilocks Effect’ of Water in Lewis-Brønsted Acid and Base Catalysis
Benedict J. Barron,a,b Wing-Tak Wong,b Pauline Chiub and King Kuok (Mimi) Hiia*
[a] Department of Chemistry, Imperial College London, South Kensington, London, SW7 2AZ, United Kingdom.
[b] Department of Chemistry, and The State Key Laboratory of Synthetic Chemistry, The University of Hong Kong, Pokfulam Road, Hong Kong.
ABSTRACT. Different catalytic protocols were evaluated in the enantioselective Pd-catalyzed aza-Michael reaction involving mono-protected phenylenediamine (PDA) derivatives. The use of these nucleophilic amines leads to the poisoning of the (monomeric) Lewis acidic catalyst, and significant competitive formation of side products were observed. In contrast, good yields and enantioselectivities can be attained by employing the Brønsted basic-Lewis acidic dimeric Pd catalyst, in combination with PDA derivatives protonated by triflic acid. In this case, the presence of the right amount of water was found to be critical for success (‘Goldilocks effect’). The results were rationalized on the basis of delicately balanced acid-base equilibria, dependent upon the nature of the catalyst and the amine.
Keywords: Asymmetric catalysis, aza-Michael reactions, deactivation, water-effect, phenylene diamine, Lewis acid, Brønsted acid and base.
Introduction
Figure 1. Dicationic diphosphine-palladium(II) chiral Lewis acid catalysts.
Dicationic Pd(II) complexes of chiral diphosphines (typically BINAP and its derivatives), 1 and 2, are a unique class of chiral Lewis acids that are able to activate a broad range of chelating Michael donors[1] and acceptors[2] towards asymmetric reactions (Fig. 1). Previously, we reported the use of these catalysts for the addition of aryl amines to Michael acceptors containing chelating 1,3-dicarbonyl groups (aza-Michael reactions), providing a route to the preparation of b-amino acid derivatives.[2a, 2b] The synthetic utility of the reaction was exemplified by the synthesis of several biologically active molecular entities containing chiral tetrhydroquinoline core structures, including torcetrapib,[3] Galipea alkaloids[4] and levonantradol[5] (Scheme 1).
Scheme 1. Complementary protocols for the Pd-catalyzed aza-Michael reactions.
One of the significant issues of any Lewis acid catalysis is catalyst poisoning by irreversible binding of nucleophilic reactants. For the Pd-catalyzed aza-Michael reactions, there are two reactions protocols that can be utilized, depending on the nature of the N-nucleophile.[6] In general, primary anilines containing electron-deficient substituents can be deployed directly as the free base using the monomeric catalyst 1 (Scheme 1, protocol A).[2a] This is the preferred method as it is operationally simple, and the precursors can be used directly as supplied. With more Lewis-basic anilines, however, catalyst inhibition can be avoided by controlled release of the amine (slow addition using a syringe pump)[7] or, more effectively, by using the dimeric catalyst precursor 2 in combination with the amine salt (Scheme 1, protocol B, ‘amine salt control’).[8] In this protocol, the Lewis basicity of the amine precursor is subdued by protonation by the Brønsted acid (triflic acid); its nucleophilicity is unmasked by the Pd-OH moiety in catalyst 2 during the catalytic process.
In our effort to extend the scope of the aza-Micheal reaction to the synthesis of more complex pharmocologically-active structures, we began to explore 1,3- and 1,4-phenylenediamine (PDA) derivatives as substrates. These diamine substrates are particularly challenging as the pKa of 1,4-PDA (6.08) is nearly an order of magnitude greater than p-anisidine (5.29), the most basic aniline substrate examined to date.[9] In this paper, we will show that excellent yields and enantioselectivities can be achieved by fine-tuning the Lewis and Brønsted acid-base chemistry.
Results and discussion
Five mono-protected diamines 3a-e containing different electron-withdrawing protecting groups (Z) were chosen in order to reduce the basicity of the nucleophile, and to prevent reaction at both nitrogens (Scheme 2). Preliminary experiments were conducted with 3a (Z = Boc) to establish the optimal reaction conditions. To ensure accurate comparison between the systems, however, THF was employed as a solvent to maintain homogeneity of the reaction mixtures, even though it is not necessarily the optimal solvent to use for every reaction (Tables S1 and S2, Supporting Information).
Scheme 2. Aza-Michael addition of 1,3- and 1,4-PDA derivatives to crotonyl-derived 4 using protocol A (see Table 1).
Using catalytic protocol A, the conjugate addition reactions of 1,3- and 1,4-PDA’s to the crotonyl-derived 4 are sluggish,[10] affording only moderate yields and enantioselectivities after two days at 45 °C (Table 1). More notably, side products were also observed in all the reaction mixtures, which were identified as the reduced Michael acceptor (6) and the oxidized Michael adduct (7), contributing to 15-25% of the mass balance, increased by elevated temperatures (Table S3 and S4, Supporting Information). The nature of the protecting group has a significant effect on the reaction outcome; the rates of reaction (monitored by 1H NMR spectroscopy) were found to increase in the order: 3b (Ac) < 3d (Piv) < 3e (Ts) < 3a (Boc) < 3c (TFA) (Fig. S1, Supporting Information).
Table 1. Asymmetric addition of PDA derivatives to 4 using protocol A.a
Entry / Amine / Yield of 5b (%) / eec(%)1 / p-3a (Z = Boc) / 58 (49) / 77
2 / p-3b (Z = Ac) / 23 (17) / 78
3 / p-3c (Z = TFA) / 61 (57) / 73
4 / p-3d (Z = Piv) / 49 (45) / 65
5 / p-3e (Z = Ts) / 39 (31) / 68
6 / m-3a (Z = Boc) / 51 (45) / 81
7 / m-3b (Z = Ac) / 43 (41) / 65
8 / m-3c (Z = TFA) / 37 (35) / 76
9 / m-3d (Z = Piv) / 39 (40) / 68
10 / m-3e (Z = Ts) / 29 (24) / 76
a Reaction conditions: Complex 1 (10 mmol, 5 mol%), 3a-e (0.2 mmol, 1 equiv.), 4 (0.3 mmol, 1.5 equiv.), THF (0.7 mL), 45 ºC, 48 h. b Calculated from 1H NMR integrations. Isolated yields are indicated in parentheses. c Determined by chiral HPLC.
The formation of these side-products suggests that the decomposition of aza-Michael products is competitive under these catalytic conditions. This was examined by subjecting an optically active (99% ee) adduct p-5a to different reaction conditions. In the absence of catalyst, optically active p-5a was stable in THF at 45 °C for 2 days (Table 2, entry 1). Upon the addition of 5 mol% of rac-1, the formation of Michael acceptor 4 and disproportionation products 6 and p-7a were detected, with some degradation in the ee of p-5a (entry 2). By lowering the reaction temperature, the decomposition can be reduced, but not totally suppressed (entry 3). No significant improvement in yield or enantioselectivity can be achieved by slow addition of the nucleophile over the course of the reaction (Table S5, Supporting Information).
Table 2. Stability of aza-Michael adduct (p-5a, 99% ee) under different conditions.a
Entry / Catalyst (mol%) / T (°C) / p-5a:4:6b / ee of 5ac (%)1 / 0 / 45 / 100 : 0 : 0 / 99
2 / 5 / 45 / 85 : 11 : 4 / 95
3 / 5 / 30 / 94 : 5 : 1 / 99
a Reaction conditions: rac-1/p-5a (0.05:1), THF (0.2 M), 30–45 ºC, 48 h. b Calculated from 1H NMR integrations. c ee of recovered p-5a, determined by chiral HPLC.
The results indicated substantial catalyst deactivation in these reactions. In our previous mechanistic work,[7] we have establish that aniline binds to diaqua complex 1 to form a 1:1 complex (8) reversibly, while the 1:2 adduct (8′) is kinetically stable and catalytically inactive (Scheme 3). With the use of the more basic PDA derivatives, a different catalyst deactivation pathway was uncovered for catalyst protocol A, involving the formation of the Pd-µ-hydroxy-µ-amido dimer (9), which was established by a series of 31P NMR spectroscopic experiments (Scheme 3 and Figure 2): Initial addition of 0.5 equivalent of p-3a to a solution of complex 1 in THF (dP +32 ppm, Figure 2a) led to the broadening of the singlet resonance as p-3a, solvent (THF) and water compete for binding to the metal center.[7] Additionally, a 31P singlet resonance signal was also observed at 28.5 ppm (Fig. 2b), accompanied by the appearance of a distinctive 1H resonance signal at ca. –3 ppm (observed in separate CDCl3 1H NMR spectroscopy experiments), which corresponds to the formation of the m-hydroxo-bridged palladium dimer 2. With the addition of ≥1 equivalent of p-3a, a distinctive ABCD coupling pattern emerged (Fig. 2c and 2d), signifying the formation of the Pd-µ-hydroxy-µ-amido dimer 9,[11] where one of the bridging hydroxo ligands in 2 is substituted by a bridging amido ligand. This appears to be a thermodynamically favorable process, as the reversal to complex 2 was only observed upon the addition of 50-200 equivalents of water (Fig. S2, Supporting information).
Scheme 3. Deactivation pathways for catalyst 1 by different anilines (X = OTf).
Figure 2. 31P NMR spectra of mixture of complex 1 and p-3a. a) Complex 1. b) Complex 1 + p-3a (0.5 eq.). c) Complex 1 + p-3a (1 eq.). d) Complex 1 + p-3a (2 eq.).
Complex 9 is catalytically inactive under reaction conditions employed in protocol A, i.e. its formation inhibits catalysis. Similar dimeric m-hydroxy-m-anilido complexes were first reported by Pregosin and co-workers[11a] and also observed by Sodeoka and co-workers in the mechanistic study of catalytic protocol B. In both cases, the complex was generated by the addition of anilines to the complex 2, which can deprotonate the aromatic amine via its Brønsted basic OH groups. In this work, we have shown that complex 9 can also be generated from the acidic complex 1 in the presence of particularly nucleophilic anilines. In the present case, the PDA derivatives are sufficiently basic to deprotonate the coordinated H2O ligand in complex 1 to generate complex 2, which then reacts with the free amine to afford complex 9 (Scheme 3). Indeed, complex 9 can also be generated from other basic aniline derivatives such as p-anisidine and N,N-dimethyl-p-phenylenediamine (Fig. 3). In each case, the formation of the dimeric complex was clearly indicated by the distinctive ABCD pattern in the 31P-{1H} NMR spectra, further supported by the observation of their distinctive M+ ion by ESI-MS (Fig. S3, Supporting information).
Figure 3. 31P NMR spectra of mixture of complex 1 and para-substituted aniline derivatives (2 equivalents): a) p-anisidine; b) N,N-dimethyl-p-phenylenediamine; c) p-3a.
In the second half of the work, we examined the utility of the second catalytic protocol (B) for the addition of PDA derivatives. First reported by Sodeoka and co-workers,[8] it is known to be amenable for aza-Michael reactions of more nucleophilic substrates, including anisidine. For this work, the triflate salts of PDA derivatives (3a-e·HOTf) were employed. These act as latent nucleophiles, which react with the Brønsted basic catalyst 2 to release the active catalyst 1 and an equivalent amount of the donor substrate.
As the PDA-triflate salts are mildly hygroscopic, they were initially prepared and used under strictly anhydrous conditions. However, the results obtained for the catalytic reactions were found to be rather erratic and irreproducible (Table S6, Supporting Information). After some investigation, we found that the presence of water is critical to ensure good catalytic activity, selectivity and reproducibility (Table S7, Supporting Information). This is further demonstrated by the reaction between 4 and m-3e·HOTf catalyzed by complex 2 in THF-d8 containing varying amounts of water (Fig. 4):[12] the presence of 6-12 equivalents of water (♦ and ▼) is optimal for catalyst turnover, affording the expected product with 93% ee. Notably, the turnover is suppressed with little to no water added (« and □), suggesting that water is needed to generate the active catalyst. Conversely, the addition of too much water (24 equivalents, ●) led to slower reaction, and reduction of ee to 88%.
Figure 4. 1H NMR reaction profile study on water effects with m-3e·HOTf, 4 and complex 2 in THF-d8. Reaction conditions: 2/m-3e•HOTf/4 (0.025:1.5:1), N-carbamate = 0.25 M in THF-d8, r.t., D2O stoichiometry:« 0 eq.,a □ 1eq., ♦ 6 eq., ▼ 12 eq., ● 24 eq. %Conversions were calculated from 1H NMR integrations and ee values were determined by chiral HPLC. a 0 equivalents of water led to polymerisation of THF-d8 after 48 h.
Table 3. Asymmetric addition of phenylenediamine triflate salts using catalyst 2.a
Entry / Amine / Yieldb (%) / eec (%)1 / p-3b / 90 (82) / 90
2 / p-3c / 85 (83) / 96
3 / p-3d / 80 (73) / 99
4 / p-3e / 90 (86) / 94
5 / m-3b / 80 (75) / 94
6 / m-3c / 84 (82) / 96
7 / m-3d / 88 (82) / 96
8 / m-3e / 91 (78) / 94
a Reaction conditions: Catalyst 2, 3a-e·HOTf, 4, H2O (0.025:1.5:1:6), [4] = 0.5 M in THF, 30 ˚C, 48 h. b Calculated from 1H NMR integrations using 1,3,5-trimethoxybenzene as an internal standard. Values in parentheses denote isolated yields. c Determined by chiral HPLC.
Subsequently, the catalysed addition of the PDA derivatives was assessed under ‘wet’ conditions in the presence of 6 equivalents of water (Table 3). In these cases, the use of the triflate salt as precursor necessitates a basic workup of the reaction mixture.[6] Another disadvantage of Protocol B is its incompatibility with acid-labile protecting groups (3a•HOTf cannot be prepared). Nevertheless, excellent yields and enantioselectivities of ≥90% can be consistently achieved, regardless of the substitution pattern of the PDA (1,3- or 1,4-) or the nature of the protecting group. In most cases, the enantiomeric excesses of the Michael adducts may be further enhanced by recrystallization to afford optically pure compounds.
The absolute stereochemistry of the major enantiomer afforded by these reactions was established by independent syntheses of (S)-10 from the aza-Michael adduct and (S)-3-aminopropanoic acid (Scheme 4). In accord with earlier studies,[3, 7] the use of (R)-BINAP catalysts favours addition on the si-face of 3 in these asymmetric processes. Specifically, the stereoselectivity remained the same between the two catalyst protocols.
Scheme 4. Independent syntheses of (S)-(+)-10. a) 1-iodo-4-nitroaniline (1 equiv.), CuI (20 mol%), K2CO3 (2.5 equiv.), DMF/H2O (50/1), 150 ºC, 42 h. 44%; b) H2, Pd/C MeOH, r.t., 2 h, 92%; c) Boc2O, THF, rt ,18 h, 84%.