Chiral phosphoramide-catalyzed, enantioselective, directed cross-aldol reactions of aldehydes

Abstract

Catalytic, enantioselective, directed cross-aldol reactions of aldehydes are described. The addition of isobutyraldehyde trichlorosilyl enolate 2 to various aldehydes in the presence of 10 mol % bisphosphoramide 4 provides aldol products in high yields with moderate to good enantioselectivities. The reaction works well with a wide range of aromatic, olefinic, and aliphatic aldehydes. Enantioselectivities are highly dependent on the electronic nature of the aldehyde substituent. Hammett studies reveal that enantioselectivity increases as aldehydes become either more electron rich or more electron poor.

The catalytic enantioselective aldol reaction has achieved “strategy level” status in organic synthesis (1–6). In this process, a new carbon-carbon bond is formed with the creation of up to two new stereogenic centers. The relative and absolute configurations of the newly created stereocenters are usually controlled through the use of chiral Lewis acids (7) or through enzymatic methods (3, 8). Chiral Lewis acid-catalyzed aldol reactions generally afford aldol products from enoxysilane derivatives of ketones or esters [directed aldol reaction (9)] in high yields with good to excellent selectivity. A few notable examples of chiral Lewis acids used in these catalytic aldol processes include bisoxazoline copper(II) complexes (10, 11) and Schiff base titanium(IV) complexes (12, 13), which have been developed by Evans et al. and Carreira et al., respectively. Recent advances also include the development of direct catalytic enantioselective aldol reactions that employ the two unmodified carbonyl components. Early reports on this type of aldol reaction using LaLi₃ tris(binaphthoxide) as a catalyst were disclosed by Shibasaki et al. (14–16). Moderate to good yields and selectivities were obtained in these studies, albeit at long reaction times. Further, Trost et al. (17, 18) have developed dinuclear zinc catalysts for direct catalytic enantioselective aldol reactions. These catalysts, which mimic class II aldolases, have been shown to catalyze aldol reactions of unmodified ketones with aldehydes in moderate to good yields with excellent levels of stereocontrol. Parallel to these successful developments is the finding of proline and its derivatives as effective catalysts for direct enantioselective aldol reactions (19–21). These simple cyclic amine catalysts are capable of mimicking class I aldolases, and they provide another approach to asymmetric aldol reactions (22, 23).

Besides these methods, the use of enzymes (3, 24) and monoclonal antibodies (8) in enantioselective aldol reactions has also been described with success. Wong et al. (24) have demonstrated the use of natural aldolase-catalyzed aldol reactions as a means to effectively and selectively synthesize highly oxygenated carbon chains, allowing a rapid access to various carbohydrates. However, broadening enzyme specificities and being able to perform enzymatic transformations on a preparative scale remain challenges for further development.

For the directed aldol reaction, silyl enol ethers derived from ketones and esters are most often used in the catalytic process. Silyl enol ethers derived from aldehydes (5) are rarely used because of their propensity for side reactions leading to undesired products (Fig. 1). For example, the aldehyde can self-condense to give homo aldol products (path a). Furthermore, dehydration of the aldol product produces an α,β-unsaturated aldehyde, which can act as a Michael acceptor (path b). Finally, the aldol adduct can undergo oligomerization or Tischenko-type processes (paths c and d).

Fig. 1.

Problems with cross-aldol reactions of aldehydes.

Despite these problems, in 1980 Heathcock et al. (25) reported the diastereoselective aldol reaction between benzaldehyde and a preformed lithium enolate derived from propanal. The aldol product is obtained in moderate yield and low diastereoselectivity, which is independent of the starting enolate geometry. In 1997, Mahrwald et al. (26–28) disclosed a different approach to directed, cross-aldol reactions of aldehydes that involves the use of an amine base and titanium tetrachloride as a Lewis acidic promoter. Enolates generated in situ by this method undergo aldolization to afford aldol products in moderate yields with high syn selectivities depending on choice of aldehyde. In the same year, Oshima et al. (29) reported the aldol additions of trichlorotitanium enolates generated by both reduction and conjugate addition from α-iodoaldehydes and allyltrimethylsilane. An improved procedure also developed by this group involves the use of geometrically defined E and Z tributoxytitanium enolates in cross-aldol additions (30). Aldol products are obtained in moderate to good yields with variable diastereoselectivity. Recently, following on their initial studies on proline-catalyzed aldol reactions, Barbas et al. (31) have reported enantioselective self-aldol reaction of acetaldehyde, albeit in low yield. Subsequently, MacMillan et al. (32) demonstrated that enantioselective direct cross-aldol reactions of aldehydes could be achieved with the use of proline as the catalyst. Aldol products are obtained in good yields with excellent enantioselectivities, but diastereoselectivities are variable, depending on the aldehyde structures.

Background

An alternative approach to catalytic enantioselective aldol reactions in general has been developed in our laboratory by the use of chiral Lewis bases and reactive trichlorosilyl enolates (33–37). In this aldol process, the chiral Lewis base (G*) is believed to bind to the silicon atom of the trichlorosilyl enolate, resulting in ionization of a chloride ion and generation of a cationic silicon species (Fig. 2) (4, 38–40). This cationic species (41) binds the aldehyde, thus generating a reactive ternary complex that undergoes reaction through a closed, six-membered-ring transition structure. The reactive silyl cation acts as an organizational center that binds and activates both the electrophile and the nucleophile under the influence of the chiral Lewis basic moiety. The aldol process is characterized by high yield, broad scope, and good to excellent selectivity.

Fig. 2.

Catalytic cycle for chiral Lewis base-catalyzed aldol addition reaction.

In a preliminary account, we reported the catalytic, enantio- and diastereoselective directed, cross-aldol reaction of aldehydes (Fig. 3) (42). The reaction provides β-hydroxyaldehydes protected as dimethyl acetals in high yields with moderate to good enantioselectivities. Unlike the proline-catalyzed direct cross-aldol reaction of aldehydes, the diastereoselectivity of the reaction is excellent and correlates well with enolate geometry. This unique feature allows access to all possible stereoisomers of the aldolate architecture by changing enolate geometry and catalyst configuration. Moreover, the reaction proceeds cleanly with no requirement for slow addition of either enolate or aldehyde.

Fig. 3.

Our first catalytic diastereo- and enantioselective directed cross-aldol reaction of aldehydes.

Herein, we report the aldol addition of the trichlorosilyl enolate derived from isobutyraldehyde to a wide range of aldehydes in the presence of a catalytic amount of a chiral bisphosphoramide. The enantioselectivity of the aldol reaction was examined with aldehydes of electronically distinct nature to gain insight into the substituent effects on the selectivity of the reaction. Finally, the scope of the reaction was investigated to evaluate its synthetic utility.

Materials and Methods

Preparation of 1-Trichlorosiloxy-1-propene (2). To a stirred solution of isobutyraldehyde (9.20 ml, 100 mmol) and hexamethylphosphoric triamide (1.72 ml, 10 mmol, 0.1 eq) in 50 ml of methylene chloride, was added, dropwise, silicon tetrachloride (23 ml, 200 mmol, 2 eq) and 2,4,6-trimethylpyridine (13.2 ml, 100 mmol, 1 eq) at 0°C. On the addition of trimethylpyridine, a white precipitate formed. The reaction mixture was allowed to warm to room temperature and was stirred under a nitrogen atmosphere for 16 h. The reaction mixture was then transferred by means of a cannula into a separate flask, leaving the white solid residue behind. The solvent was removed by vacuum distillation (200 mmHg at 25°C; 1 mmHg = 133 Pa). A second distillation of the residue at 38°C, 20 mmHg, afforded 10.75 g (52%) of 2 as a clear, colorless liquid. Data for 2: bp 38°C (20 mmHg); ¹H NMR (400 MHz, CDCl₃) δ 6.14 (sept, J = 1.4, 1 H, HC (1)), 1.66 (d, J = 0.97, 3 H, H₃C (3)), 1.61 (d, J = 0.97, 3 H, H₃C (3)); ¹³C NMR (500 MHz, CDCl₃) δ 129.40 (C (1)), 120.35 (C (2)), 19.00 (C (3)), 15.15 (C (3)); MS (electron impact) m/z 205 (M⁺, 25), 204 (52), 191 (100), 169 (34), 132 (37).

Procedures for the preparation and characterization of enolate 2 are also published as supporting information on the PNAS web site.

General Procedure for the Chiral Phosphoramide-Catalyzed Cross-Aldol Reactions of Aldehydes. Reaction of (R)-(2,2-dimethoxy-1,1-dimethylethyl)benzenemethanol (5a) is described as an example. To a stirred solution of 4 (84.3 mg, 0.1 mmol, 0.1 eq) in 4 ml of chloroform/methylene chloride (4/1) at –78°C was added dropwise enolate 2 (226.1 mg, 1.1 mmol, 1.1 eq). Five minutes later, freshly distilled benzaldehyde (102 μl, 1 mmol) was added dropwise. The reaction mixture was stirred at –78°C under N₂ for 8 h. Dry methanol (15 ml) was then added dropwise at –78°C, and the reaction mixture was stirred for 45 min. The solution was allowed to warm to room temperature (30 min) and was then quickly poured into a rapidly stirred, cold, aqueous sodium bicarbonate solution (25 ml). The resulting mixture was stirred at room temperature for 2 h. A cloudy white solution was observed during this period. The mixture was filtered through Celite and the aqueous layer was separated and extracted with methylene chloride (three times, 25 ml each). The combined organic layers were dried over sodium sulfate, filtered, and concentrated in vacuo. Silica gel column chromatography (hexanes/ethyl acetate, 5/1) followed by bulb-to-bulb distillation afforded 192.6 mg (86%) of 5a as a clear, colorless oil. Data for 5a: bp 140°C (0.02 mmHg); ¹H NMR (400 MHz, CDCl₃) δ 7.23–7.33 (m, 5 H, H-Aryl), 4.76 (d, J = 2.2, 1 H, HC (5)), 4.05 (s, 1 H, H(C (2)), 3.87 (d, J = 2.2, 1 H, OH), 3.58 (s, 3 H, H₃C (3)), 3.56 (s, 3 H, H₃C (3)), 0.92 (s, 3 H, H₃C (4)), 0.76 (s, 3 H, H₃C (4)); ¹³C NMR (400 MHz, CDCl₃) δ 141.13 (C (6)), 128.01 (C (7)), 127.47 (C (9)), 127.14 (C (8)), 114.02 (C (2)), 77.95 (C (5)), 58.82 (C (3)), 58.67 (C (3)), 43.62 (C (1)), 21.13 (C (4)), 17.04 (C (4)); IR (CHCl₃) 3,629 (w), 3,479 (br), 3,009 (s), 2,942 (s), 2,836 (m), 1,471 (m), 1,453 (m), 1,390 (w), 1,339 (w), 1,237 (w), 1,190 (m), 1,104 (s), 1,070 (s), 1,016 (s), 955 (w), 705 (m) cm^–1; MS (field ionization) m/z 224 (M⁺, 0.4), 192 (M⁺ – MeOH, 4), 106 (C₆H₅CHO, 33), 86 ((CH₃)₂CCHOMe, 100); [α]_D–4.56 (c = 0.74, EtOH); TLC R_f 0.30 (hexane/EtOAc, 5/1) [silica gel, visualization with 2,4-dinitrophenylhydrazine (DNP)]; supercritical-fluid chromatography (R)-5a, t_R 2.78 min (70.2%); (S)-5a, t_R 3.19 min (29.8%) (column: OD, 5% MeOH, pressure 150 psi, flow 3.0 ml/min). Analysis. Calculated for C₁₃H₂₀O₃ (224.30): C, 69.61; H, 8.99; found: C, 69.33; H, 9.02.

Procedures for the preparation and characterization of aldol products and acylated aldol products, as well as representative procedures for the aldol additions, are published as supporting information.

Results

Preparation of Trichlorosilyl Enolate 2. The synthesis of trichlorosilyl enolates derived from ketones was previously accomplished by metathesis of trimethylsilyl enol ethers with silicon tetrachloride (SiCl₄) in the presence of a catalytic amount of mercury(II) acetate or palladium(II) acetate (43). In the presence of 5 mol % of either Hg(OAc)₂ or Pd(OAc)₂, the metathetical reaction of trimethyl-[2-methyl-(propenyloxy)]silane 1 with SiCl₄ proceeded very slowly at room temperature to afford only trace amounts of trichlorosilyl enolate 2 according to ¹H NMR analysis (Fig. 4). Increasing the catalyst loading or reaction concentration did not provide satisfactory results. The attenuated reactivity of enol ether 1 is presumably due to steric hindrance at the β-carbon of the enoxysilane. Another approach to aldehyde-derived trichlorosilyl enolates is the cleavage of the corresponding trimethylsilyl enol ethers with methyllithium, followed by trapping of the resulting lithium enolates with SiCl₄ (42). Although this method could be used to prepare enolate 2, a much simpler, more direct synthesis of enolate 2 was sought.

Fig. 4.

Initial preparation of isobutyraldehyde trichlorosilyl enolate 2.

Previous studies in this laboratory have shown that enol silylation of ketones with SiCl₄ and an amine base is catalyzed by hexamethylphosphoric triamide (HMPA) (43). We were pleased to find that the reaction of isobutyraldehyde with SiCl₄ and triethylamine in the presence of 10 mol % HMPA in methylene chloride proceeded smoothly at room temperature to afford trichlorosilyl enolate 2 (Fig. 5). Unfortunately, the formation of triethylamine hydrochloride under the reaction conditions created difficulty in product isolation. However, the use of 2,4,6-trimethylpyridine under similar reaction conditions avoided this problem and afforded enolate 2 in modest yield after distillation. This reaction provides a simple single-step procedure for preparation of the enolate directly from isobutyraldehyde. By using this method, trichlorosilyl enol ether 2 can be synthesized on a multigram scale and can be stored for months under inert atmosphere without decomposition or hydrolysis.

Fig. 5.

Improved preparation of isobutyraldehyde trichlorosilyl enolate 2.

Chiral Phosphoramide-Catalyzed Aldol Additions. The reactivity of enolate 2 toward benzaldehyde (3a) in the presence of 10 mol % the dimeric phosphoramide (R,R)-4 was examined (Table 1). Gratifyingly, the reaction proceeded smoothly at –78°C to afford the desired product 5a within 8 h in excellent yield albeit rather moderate selectivity. The choice of solvent for the reaction was based on the results of solvent screening studies previously performed in these laboratories (See ref. 42. Other solvents, such as toluene, chloroform, methylene chloride, and diethyl ether were used. Chloroform/methylene chloride, 4/1, gave the best yields and selectivities.). Not surprisingly, in the absence of the bisphosphoramide, the addition of enolate 2 to benzaldehyde did not yield the aldol product after 5 h at –74°C. Directly monitoring the reaction through the ReactIR (ReactIR 1000 fitted with a ⅝-inch DiComp Probe, running software version 2.1a, from ASI Applied Systems, Millersville, MD) indicated that the benzaldehyde persisted unchanged during the 5-h period. The background reaction was first quenched with methanol at –74°C and then worked up with NaHCO₃. Spectroscopic (¹H NMR) analysis of the crude reaction mixture revealed that virtually no aldol product was formed. Subsequent purification of the crude material allowed for almost quantitative recovery of benzaldehyde in the form of its dimethyl acetal.

Table 1.

Enantioselective aldol addition of 2

Entry	R	Product	Time, h	Yield,* %	er†
1	C6H4	5a	8	86	70.0/30.0
2	4-MeC6H4	5b	12	90	73.0/27.0
3	4-MeOC6H4	5c	20	92	75.5/24.5
4	3,4,5-(MeO)3C6H2	5d	26	80	87.5/12.5
5	4-ClC6H4	5e	8	85	89.0/11.0
6	4-CF3C6H4	5f	8	86	90.0/11.0
7	4-NO2C6H4	5g	8	89	91.0/9.0
8	2-Naphthyl	5h	12	90	83.0/17.0
9	(E)-Cinnamyl	5i	12	90	67.5/32.5
10	Phenylpropargyl	5j	12	85	81.5/18.5
11	1-Propenyl	5k	15	82	56.0/42.0‡
12	n-Butyl	5l	30c	80	91.0/9.0‡

All reactions were run at -78°C except valeraldehyde at -20°C.

^*Yield of analytically pure materials.

^†Enantiomeric ratio determined by chiral stationary phase-supercritical fluid chromatography on Daicel Chiralpak, OD, AS, and AD columns.

^‡Enantiomeric excesses were determined on the corresponding benzoate products.

To further examine the reactivity and the selectivity of the aldol reaction, a wide range of aldehydes with electronically distinct character was surveyed (Table 1). In general, moderate to good selectivity was obtained for both electron-rich and electron-poor aromatic aldehydes (entries 2–7). Electron-poor aldehydes were generally more reactive than electron-rich aldehydes and required shorter reaction times (entries 5–7 vs. 2–4). Moreover, the rates of aldolization with electron-poor aldehydes were faster than those with electron-rich aldehydes. These results implied that aldolization is the rate-determining step.‡ 2-Naphthaldehyde gave an excellent yield but only moderate selectivity (entry 8). Olefinic aldehydes afforded products in good yield with selectivity lower than those obtained from aromatic aldehydes (entries 9–11 vs. 2–7). Interestingly, an aliphatic aldehyde still reacted although elevated temperature was required, affording product in good yield and selectivity (entry 12).

Detailed examination of the enantiomeric ratios for electron-rich and electron-poor aldehydes revealed that two distinct trends in selectivity exist. Taking benzaldehyde as a reference point, the selectivity increased as the aldehydes became more electron rich (Table 1, entries 1–4). Interestingly, the selectivity also increased as the aldehydes became more electron poor. However, overall greater selectivity was observed in these latter cases (entries 5–7).

To gain insight into this surprising phenomenon, a Hammett plot was constructed (44, 45). A reasonable correlation between the selectivity and the electronic effects of substituents on aldehydes was obtained (Fig. 6). Sigma (σ) and inductive sigma (σ_i) values were used for electron-rich and electron-poor aldehydes, respectively (46–48). It was noted that if σ_i values were not used, 4-chloro substituent would be the only scattered data point. This was expected because the chloro substituent is known to possess both inductive and resonance effects, which are opposed to one another, depending on the nature of the reaction (46). To avoid this complication, σ_i values were used for the electron-poor aldehyde series. Interestingly, the Hammett study showed that substituents on the aromatic ring have a dramatic effect on the selectivity of the reaction. Similar ρ values (but of opposite sign) were obtained from both electron-rich and electron-poor substrates, indicating that the sensitivity of the aldol reaction to the electronic effects in both cases is similar. Previous studies in this laboratory have shown that electron-rich aldehydes tend to be more enantio- and diastereoselective in the phosphoramide-catalyzed aldol addition using ketone-derived trichlorosilyl enolates (49). These studies have also shown that there is no apparent trend in selectivity for electron-poor aldehydes.

Fig. 6.

Substituent effects on enantiomeric ratio (er).

Most striking is the change in slope in the Hammett plot. A break in the Hammett plot can be interpreted as either a change in the stereochemistry-determining step as substrates change from electron-rich to electron-poor aldehydes, or a change in factors that control the two selectivity regimes (50–52). This implication poses a mechanistically interesting question as to whether the configurations of the hydroxyl-bearing methine carbons in the aldol products at both ends of the electronic spectrum are the same. To resolve this stereochemical issue, the absolute configurations of aldol products 5d and 5f were determined. Aldol adducts 5d and 5f were acylated with 4-bromobenzoyl chloride, and the resulting products 5m and 5n were then recrystallized from hexanes and isopropyl ether to near enantiomeric purity (Fig. 7). In both cases, the major enantiomers were isolated and assessed to be about 98/2 enantiomeric ratio according to analysis by chiral stationary phase–supercritical fluid chromatography (CSP-SFC). Single-crystal x-ray structures of 5m and 5n were obtained from these samples. The x-ray data are published as supporting information. (The crystallographic coordinates of 5m and 5n have been deposited with the Cambridge Crystallographic Data Centre, deposition numbers 223773 and 223774, respectively. These data can be obtained free of charge at www.ccdc.cam.ac.uk/conts/retrieving.html.) The absolute configurations of both products were shown to be R from anomalous dispersion analysis. Thus, the aldol products from both electron-rich and electron-poor aldehydes had the same configuration.

Fig. 7.

Synthesis of compounds for establishing absolute configuration of aldol products by x-ray analysis. DMAP, dimethylaminopyridine.

Discussion

An operationally simple method to prepare trichlorosilyl enolate 2 directly from isobutyraldehyde under mild conditions has been developed. This procedure is best suited for synthesizing symmetrical trichlorosilyl enolates, wherein the geometry is of no concern.

Notably, the reactivity of trichlorosilyl enolate 2 toward aldehydes was less than that of its analogous propanal-derived trichlorosilyl enolate. Higher catalyst loadings and longer reaction times compared with those obtained from previous studies were needed (42). This attenuated reactivity is presumably due to steric encumbrance created by an additional methyl group at the α-carbon of enolate 2.

The chiral phosphoramide-catalyzed, directed cross-aldol reaction of aldehydes is quite general. Aromatic, olefinic, and aliphatic aldehydes all proceeded smoothly to afford products in high yields with moderate to good selectivity. The reaction provides direct access to β-hydroxyaldehydes protected as their dimethyl acetals in a single operation (Fig. 8, path a). In contrast to other procedures such as allylation and aldol addition of a ketene acetal to aldehydes, oxidation or reduction and an additional step for protection are required to access the same synthetic target (paths b and c). The preparation of the β-hydroxyaldehyde moieties described herein requires minimal functional group manipulation.

Fig. 8.

Synthetic advantages of directed cross-aldol reactions of aldehydes.

The high yields observed in the Lewis base-catalyzed, directed cross-aldol reaction of aldehydes attest to a unique mechanistic feature of this reaction. The formation of a chlorohydrin intermediate, i (which was observed spectroscopically at low temperature), as proposed previously, is crucial to circumvent potential side reactions (Fig. 9), thus allowing for the desired aldol products to form in excellent yields.

Fig. 9.

In situ trapping of aldolate and formation of β-hydroxyacetal product.

Most interestingly, two distinct trends in selectivity for electron-rich and electron-poor aldehydes highlight the mechanistically complex nature of the Lewis base-catalyzed, directed cross-aldol reaction of aldehydes. These results suggest that factors controlling selectivity for electron-rich aldehydes must be different from those for electron-poor aldehydes. Our original hypothesis for the divergence in selectivity involved a change in the facial orientation of the aldehyde with respect to the enolate upon changing the electronic nature of the aldehyde. Consequently, one would expect the absolute configurations of aldol products at both ends of the electronic spectrum to be different. However, x-ray analysis disproved this hypothesis, as both electron-rich and electron-poor aldehydes led to aldol products with the same absolute configuration.

Other possibilities for the divergence of selectivity may also be considered. Among them, the most reasonable are (i) a change in the stereochemistry-determining step, (ii) a change in factors that influence selectivity, and (iii) a change in the rate-determining step. Alternatively, this behavior may also be attributable to a change in the dominance of enthalpic or entropic contributions to the activation free energies of the different processes. These hypotheses can be probed experimentally, and the execution and interpretation of these experiments are the subject of a forthcoming paper.

Conclusions

A simple method for the synthesis of trichlorosilyl enol ether 2 has been developed. The addition of the trichlorosilyl enolate of isobutyraldehyde (2) to a variety of aldehydes in the presence of 10 mol % phosphoramide (R,R)-4 afforded β-hydroxyaldehydes protected as dimethyl acetals in high yields with moderate to high enantiomeric purity. This method provides direct access to a latent 1,3-diol motif or its synthetic equivalent containing a quaternary center with minimal functional group manipulation. The experimental results highlight strong electronic effects of substituents on the enantioselectivity of the reaction. Mechanistic investigations of the origin of the electronic effects are needed, as well as the development of tandem addition processes.