Aza-Rubottom Oxidation: Synthetic Access to Primary α-Aminoketones

Abstract

An aza analogue of the Rubottom oxidation is reported. This facile transformation takes place at ambient temperature and directly converts silyl enol ethers to the corresponding primary α-aminoketones. The use of hexafluoro isopropanol (HFIP) as the solvent is essential for the success of this reaction. Overall this process is well-suited for the aza-functionalization and derivatization of complex organic molecules.

Graphical Abstract

The alpha-aminoketone moiety is commonly found in biological molecules, natural products and active pharmaceutical ingredients. Despite their apparent importance, synthetic access to these compounds is not always straightforward.¹ This is especially evident when the desired amino ketones have a primary amino group attached to a fully substituted α-carbon atom. The most common strategy for synthesizing these compounds involves a two-step approach, in which either an azido,² a nitro,³ or a hydroxylamino group⁴ is first installed at the α-position. A subsequent hydrogenation, usually catalyzed by a transition metal such as Pd, Pt or Raney Ni, is then carried out to obtain the corresponding primary α-aminoketone (Figure 1, A). Aside from the added steps, this approach also has limitations such as the lack of chemoselectivity during the reduction and the potential instability of the azido- or nitro- intermediates.

Figure 1.

Existing methods for the synthesis of α-amino-ketones and the proposed “aza-Rubottom oxidation”

Significant efforts have been devoted into making α-hydrazinyl compounds from ketones and azodicarbonyl compounds such as DEAD under basic or organocatalytic conditions (Figure 1, B).⁵ Although this type of α-hydrazination reaction is generally high-yielding and can be rendered enantioselective, it is far from trivial to cleave the N-N bond and reveal the desired primary amino ketones. An ideal reaction should be able to directly install the primary amino group without the need to perform an additional reduction or N-deprotection. A few such examples have been reported in the literature.⁶ The common strategy is to trap the enolates generated from the corresponding ketones with an electrophilic nitrogen source. However, good yields can only be obtained with ketones that can form stable enolates, such as malonates, phenyl acetates and phenylacetonitriles. Moreover, the required aminating reagents (i.e. chloramine, O-phosphinylhydroxylamine) are usually very volatile and their use is limited due to safety concerns (i.e., toxicity). One unique approach developed by the Wirth group uses a hypervalent iodine reagent to facilitate an intramolecular umpolung reaction to access α-aminoketones, although the amino group must be fully substituted.⁷ Other noteworthy approaches to access secondary and tertiary α-amino carbonyl compounds include several types of copper-catalyzed coupling reactions.⁸

Contrary to the generally difficult α-amination reactions, the α-hydroxylation of ketones is a well-established transformation that today is considered to be trivial. Two general approaches are available, namely the trapping of an in-situ generated enolate with Davis oxaziridine⁹ and the Rubottom oxidation in which silyl enol ethers are oxidized with either peroxyacids, peroxides or dioxiranes.¹⁰ These two approaches have wide applicability, and both have been successfully used numerous times, even in the synthesis of extremely complex natural products.¹¹

While limited attempts have been made to mimic the first approach (i.e. the aforementioned trapping of enolates with chloramine or O-phosphinylhydroxylamine), to the best of our knowledge, no systematic studies have been dedicated to developing an amino analogue of the Rubottom oxidation (AKA “Aza-Rubottom oxidation”). Herein, we report our findings in the development of such a transformation.

The Rubottom oxidation of silyl enol ethers goes through an epoxide intermediate.^10b Presumably, an aza-Rubottom oxidation will go through the corresponding (NH)-aziridine intermediate (Figure 1, C). ^12,13 While catalyst-free epoxidation of olefins can be easily achieved with either peracids or dioxiranes, the prevailing wisdom is that NH-aziridination of olefins requires the presence of a transition metal (TM) catalyst.¹⁴

During our ongoing studies of developing enantioselective NH-aziridination reactions, we were surprised to discover that when HFIP is used as the solvent, a noticeable background reaction is present when the olefin substrates are very electron rich. This is especially evident when a tetra-substituted olefin is used, which can be fully converted to the corresponding NH-aziridine with no transition metal catalyst, albeit at a significantly slower rate. We ruled out the possibility of trace metal contamination in the HFIP by conducting a control experiment using redistilled HFIP, which gave the same result. This background reaction is negligible unless the olefins are very electron-rich (i.e., tetra-alkyl substituted).

Realizing that silyl enol ethers are even more electron-rich than fully carbon-substituted olefins, we were curious to see whether they could be aziridinated and in-situ converted to the corresponding α-amino ketones without the use of a transition metal catalyst.¹⁵

We started our investigation using silyl enol ether 1 and several aminating reagents (Table 1, 2–6) to determine the optimal reaction conditions (Table 1). Unsurprisingly, the reaction could proceed under virtually the same conditions used in the NH-aziridination reaction^14b and give the desired α-aminoketone 1a in good yield (Table 1, entry 1).

Table 1.

Optimization of Reaction Conditions.

Entry[a]	XONH2	Catalyst	Additive	Solvent	Yield (%)
1	2 (1.2 equiv)	Rh2(esp)2	pyridine (1.2 equiv)	HFIP	66%
2	2 (2 equiv)	none	pyridine (4 equiv)	HFIP	0b
3	2 (2 equiv)	none	K3PO4 (2 equiv)	HFIP	Traceb
4	2 (1.5 equiv)	none	Cs2CO3 (3 equiv)	HFIP	41%
5	2 (1.5 equiv)	none	Cs2CO3 (3 equiv)	TFE	Trace
6	2 (1.5 equiv)	none	Cs2CO3 (3 equiv)	MeCN	0
7	2 (1.5 equiv)	none	Cs2CO3 (3 equiv)	DCM	0
8	3 (1.5 equiv)	none	none	HFIP	47%
9	3 (1.2 equiv)	none	iPr2NEt (1.5 equiv)	HFIP	85%
10	2 (1.2 equiv)	none	iPr2NEt (1.5 equiv)	HFIP	Trace
11	4 (1.5 equiv)	none	iPr2NEt (3 equiv)	HFIP	17%
12	5 (1.5 equiv)	none	iPr2NEt (3 equiv)	HFIP	10%
13	6 (1.2 equiv)	none	iPr2NEt (1.5 equiv)	HFIP	0
14	3 (1.2 equiv)	none	iPr2NEt (1.5 equiv)	TFE	Trace
15	3 (1.2 equiv)	none	iPr2NEt (1.5 equiv)	MeCN	0
16	3 (1.2 equiv)	none	iPr2NEt (1.5 equiv)	DCM	0

^[a]Optimization reactions were conducted on a 0.2 mmol scale. Aminating reagent and additives were mixed in 1 mL solvent and stirred at rt for 10 min before the silyl enol ether was added to the mixture. Reactions were stopped after 16 h, and the amine product was converted to its HCl salt after isolation.

^[b]The silyl ether 1 decomposed to the ketone.

To facilitate the characterization and isolation of the product, we converted the free amine to an HCl salt 1b, which was found to be more stable. The aminating reaction did not proceed when the Rh catalyst was removed (Table 1, entry 2). This is likely due to the desilylation of 1 caused by the acidic reaction mixture of HOSA and HFIP which outcompetes the amination reaction in the absence of a catalyst. Indeed, when a stronger base was used, a decent yield of 1b was obtained (Table 1, entry 4).

Next, we examined several different electrophilic aminating reagents. Aminating reagents 2–5 can be considered as the nitrogen analogues of peracids, and we hypothesized that NH-oxaziridine 6, the aza analogue of dioxiranes, may behave like an aminating agent and would transfer the NH unit. The results showed that DPH (3) is the best aminating reagent (Table 1, entry 9), and it gives a better yield than the Rh-catalyzed reaction (Table 1, entry 9 vs 1). Buffering with a base is necessary to minimize desilylation and increase the yield of the desired product (Table 1, entry 9 vs entry 8). NH-Oxaziridine 6 failed to react (Table 1, entry 13), indicating that the analogy between an NH-oxaziridine and the corresponding dioxirane is merely structural and is not reflected in their reactivity. We also noticed an extreme case of solvent effect in this reaction: HFIP is absolutely essential for the success of this reaction (Table 1, entry 5–7, 14–16), and the mixing of co-solvents is detrimental to the success of this transformation.

Next, we explored the scope of the reaction (Scheme 1). Substrates bearing one phenyl group and various alkyl substituents gave good yields under the optimized conditions (Scheme 1, 1b, 7b-9b). It should be noted that an allyl group was unaffected under the reaction conditions (Scheme 1, 7b), demonstrating the reaction’s chemoselectivity, in which a more electron-rich π bond is preferentially oxidized. When R¹ was a substituted phenyl group, we observed a noticeable electronic effect in which electron-rich substrates gave higher yields than electron-poor ones (Scheme 1, 10b vs 12b). Heterocycles were tolerated in these reactions and substrates bearing thiophene, imidazole, and benzofuran rings gave satisfactory yields (Scheme 1, 13b–15b). The reaction also worked well when R¹, R² and R³ are all alkyl groups (Scheme 1, 17b and 18b). Finally, an indomethacin-derived substrate 19 was successfully aminated to furnish 19b, demonstrating this transformation’s remarkable functional group tolerance and its potential application in late-stage functionalization of medicinal chemistry intermediates. It should be emphasized that most of these products have been previously unknown.

Scheme 1. Transition Metal-Free Aza-Rubottom Oxidation.

[a] Reactions were conducted on a 1 mmol scale. DPH (1.2 equiv), diisopropylethylamine (1.5 equiv) were dissolved in 5 mL of HFIP. The silyl enol ether was added at room temperature with rigorous stirring. The purified free amines were treated with a solution of HCl in diethyl ether to obtain the corresponding HCl salts. [b] The free amine rapidly decomposes when concentrated.

However, when more strongly electron-withdrawing substituents are present in the silyl enol ethers (20–22), the reactions failed to proceed. For example, an ester group is sufficient to suppress the amination of silyl enol ether via either the inductive effect (20) or conjugation (21). In the case of 22, we suspect the in-situ protonation of the pyridinyl nitrogen in HFIP (pKa ~ 9.3) severely reduced the reactivity of the silyl enol ether to prevent the reaction from taking place. We also found that the less stable silyl ketene acetals will undergo fast desilylation under the reaction conditions, which severely reduce the yield.¹⁶

We hypothesized that in the case of less reactive silyl enol ethers, a more reactive nitrogen source is needed. Naturally, we decided to use transition metal catalysis, hoping that the more reactive metal-nitrenoid complexes can overcome the energy barrier and will aminate even these less electron-rich substrates.

The reactions worked as predicted, and primary α-aminoketones bearing electron-withdrawing substituents were obtained in good to excellent yields (Scheme 2).

Scheme 2. Transition Metal-Catalyzed Aza-Rubottom Oxidation.

[a] Reactions were conducted on a 1 mmol scale. HOSA (1.2 equiv), pyridine (2.4 equiv) were dissolved in 5 mL of HFIP. The silyl enol ether was added at room temperature with rigorous stirring, followed by the metal catalyst. The purified free amines were treated with a solution of HCl in diethyl ether to obtain the corresponding HCl salts.

Both Cu- and Rh-complexes can catalyze this transformation, although Rh is significantly more efficient (Scheme 2, 21b). Silyl enol ethers with carboxylate substituents (Scheme 2, 23b and 24b) as well as electron-deficient heterocycles (Scheme 2, 22b) were successfully aminated with the addition of only 1 mol% Rh₂(esp)₂ catalyst, and the reaction reaches completion within 6 h at room temperature. It should be noted that very recently, Professor E. J. Corey and coworker reported a similar Rh-catalyzed reaction in the context of the synthesis of ketamine analogues.¹⁷

Like the un-catalyzed reaction, the transition metal-catalyzed aza-Rubottom oxidation also requires the use of HFIP as the solvent for effective conversion. Moreover, the reaction becomes sluggish with the addition of a co-solvent. This echoes the significant rate acceleration effect by fluorinated solvents that we have observed in Rh-catalyzed NH-aziridination of unactivated olefins. Intrigued by this observation, we decided to search the literature for some insights on the mechanism of this transformation.

Although the consensus agrees that the rate acceleration by HFIP is most likely caused by hydrogen bond interactions,¹⁸ to the best of our knowledge no detailed mechanistic studies have been carried out on the effect of fluorinated solvents in olefin aziridination reactions. However, an outstanding study on the rate acceleration by HFIP in the context of olefin epoxidation was reported by the Berkessel group in 2006.¹⁹ In this study, it is reported that HFIP enables the epoxidation of olefins by hydrogen peroxide in the absence of additional catalysts. In the subsequent quantum-chemical investigation, Berkessel et. al. proposed that higher-order aggregates of HFIP molecules are responsible for the rate acceleration.

A similar mechanism is likely involved in the aza-Rubottom oxidation, in which higher-order solvent aggregates formed from multiple HFIP molecules activates the hydroxylamine-derived aminating reagent via cooperative hydrogen-bonding interactions (Scheme 3). This explains the detrimental effect caused by a co-solvent, since it will exponentially affect the concentration of the active HFIP aggregates.

Scheme 3. Mechanistic Hypothesis.

An intriguing implication based on this hypothesis is that an enantioselective version of this reaction can be potentially achieved with a chiral nonracemic hydrogen-bond donor catalyst that activates achiral hydroxylamines. This could be analogous to what has been reported for asymmetric epoxidation catalyzed by organocatalysts.²⁰

In summary, we have developed two sets of conditions for the synthesis of primary α-aminoketones from silyl enol ethers. Electron-rich substrates can be α-aminated without transition metal catalysis, while substrates bearing electron-withdrawing substituents can undergo α-amination with the help of Rh or Cu catalysts. Solvent aggregates formed from multiple HFIP molecules are likely responsible for the observed reactivity. We hope these aza-Rubottom reactions will become valuble synthetic tools for organic synthesis.