Silver(I) as a widely applicable, homogeneous catalyst for aerobic oxidation of aldehydes toward carboxylic acids in water-"silver mirror": From stoichiometric to catalytic.

The first example of a homogeneous silver(I)-catalyzed aerobic oxidation of aldehydes in water is reported. More than 50 examples of different aliphatic and aromatic aldehydes, including natural products, were tested, and all of them successfully underwent aerobic oxidation to give the corresponding carboxylic acids in extremely high yields. The reaction conditions are very mild and greener, requiring only a very low silver(I) catalyst loading, using atmospheric oxygen as the oxidant and water as the solvent, and allowing gram-scale oxidation with only 2 mg of our catalyst. Chromatography is completely unnecessary for purification in most cases.

Oxidation is a central task for organic chemists to achieve conversion of different organic compounds. Among them, oxidation of aldehydes to give carboxylic acids is one of the most well-known and most frequently used methodologies (, ), for example, by stoichiometrically using the Cr(IV)-based Jones reagent (, ), the Ag(I)-based Tollen’s reaction (), the Cu(II)-based Fehling’s reaction (), and the permanganate reagents (). Although it has long been known that aldehydes are very prone to oxidation, methods to achieve a highly efficient and clean transformation of aldehydes to carboxylic acids under mild and greener conditions are still scarce. Even today, most such oxidations still require stoichiometric amounts of hazardous oxidants (–) and often take place in harmful solvents.

With its natural abundance and inherent greener characteristics, water has been a desirable solvent for chemists (–). Although biological oxidations in water using enzymes or microorganisms are well recognized (–), it was only in 2000 that Sheldon established an aqueous-phase homogeneous catalytic aerobic oxidation methodology (, ). Yet, the method still requires a precious metal (palladium), a high pressure (30 bar), and a large amount of additive (TEMPO). In 2008, Tian et al. reported a heterogeneous catalytic aqueous-phase oxidation of aldehydes using silver(I)/copper(II) oxide (), but the method suffers from a high catalyst loading, a very limited substrate scope, and side reactions. In 2009, Yoshida and co-workers reported a water-soluble N-heterocyclic carbene (NHC)–catalyzed oxidation of aldehyde by oxygen (). However, this method still requires the reaction solvent to be a mixture of N,N′-dimethylformamide/H2O in 10:1 ratio, which is far from a complete water-phase oxidation. Recently, in 2014, Han and co-workers reported a multifunctional utilization of silver-NHC complex as catalyst to achieve different oxidation of alcohol (), but the method still relies on organic solvent and anhydrous conditions. Here, we wish to report a highly efficient, widely applicable, homogeneous silver(I)-catalyzed aerobic oxidation of a wide range of aldehydes using only water as solvent, and performed under atmospheric pressure with oxygen gas or oxygen in the air as the oxidant under mild conditions (Fig. 1).

Fig. 1

Highlights of our aerobic oxidation.

We began our investigation by introducing various silver(I) salts or complexes to benzaldehyde as a standard in air at atmospheric pressure without a balloon in a sealed tube (Table 1). We found the efficiency of the transformation to be strongly affected by the presence of inorganic salt (for example, entries 1 and 2). When the reaction was carried out without oxygen, a very low yield was obtained (entry 3), reflecting a stoichiometric aldehyde oxidation. The anions of the salt were then tested. Besides formate, only fluoride and tetrafluoroborate provided the oxidation product (entries 4 to 7). Considering that tetrafluoroborate might undergo hydrolysis to give fluoride in situ, fluoride was thus chosen as the standard anion to conduct further investigations. Upon examining the cation, surprisingly, it seemed to be the only one enabling the oxidation (entries 8 to 11). Sodium fluoride was therefore selected as the salt for optimizing the conditions. We then tested different ligands (entries 12 to 14) and found that the combination of chelating bipyridine as a ligand and a noncoordinating PF6− as the counter-ion (entry 14) achieved a quantitative yield of the corresponding oxidation. Switching from air to oxygen gas under the same atmospheric pressure also led to a quantitative isolated yield (entry 15). As a control experiment, only a trace amount of the product was detected in the absence of the silver catalyst.

Table 1

Investigation of reaction conditions.

A series of common aldehydes, including both aliphatic and aromatic examples with different functional groups, was then chosen to conduct the scope investigation with this catalytic system (Table 2). Besides benzaldehyde, which gave a quantitative yield (entry 1, compound 1), aliphatic 1-octanal also gave a quantitative yield of the corresponding acid (entry 2, compound 38). Hydrocinnamaldehyde and 1-naphthaldehyde gave very good yields of 86 and 88% (entries 3 and 4, compounds 49 and 4), respectively. With 4-fluorobenzaldehyde, the reaction only gave a 34% yield (entry 5, compound 16), whereas 4-chlorobenzaldehyde led to a 100% recovery of the starting material (entry 6, compound 18). Surprisingly, for the unsaturated cinnamaldehyde, even with all the starting material consumed, the reaction still did not give any desired product (entry 7, compound 47), whereas the unconjugated 4-allyloxy benzaldehyde led to a low conversion of the starting material and no desired product (entry 8, compound 12). p-Anisaldehyde and piperonal, bearing additional oxygen-based functional groups, also led to 100% recovery of the starting material (entries 9 and 10, compounds 6 and 5). We postulated that those substrates with inferior reactivity might be caused by the competing coordination of oxygen, nitrogen, C=C double bond, and so on, to the Lewis acidic silver(I) center within the same molecule as the aldehyde carbonyl. Thus, we rationalized that a stronger coordinating ligand may be required to release the Ag(I) center from such coordination.

Table 2

Investigation of substrate scope.

Using piperonal, unreactive under the above conditions, as the model substrate, phosphorus-based and NHC ligands were examined (Table 3). Unless otherwise noted, all experiments were carried out in house-light conditions, without light sheltering. With [(CF3)2CHO]3P, a very electron-poor ligand, we only obtained a 21% yield. Furthermore, some decomposition of the piperonal’s formacetal structure was observed (entry 2). The combination of AgPF6 with more electron-rich trifurylphosphine gave a good 66% nuclear magnetic resonance yield (entry 3); however, some decomposition (ca. 15%) of the acetal was still observed. The catalyst generated from AgPF6 and the NHC ligand IPr gave a much lower yield (entry 4). To our surprise, when we switched AgPF6 to Ag2O, an almost quantitative yield was obtained (entry 5). Isolation of the product from the reaction mixture was very easy: The aqueous reaction mixture was simply washed with common non–water-mixable organic solvent and then acidified, followed by extraction using diethyl ether. Without needing to perform flash chromatography, the product with an extremely high purity level was obtained. Meanwhile, a 50% yield was achieved when only 0.5 equivalent of the base was added, indicating that the presence of base is necessary to drive the reaction. The controlled experiment was then conducted to test how the reaction proceeds in the absence of oxygen. Surprisingly, with the reaction conducted with argon at 1 bar and using ordinary distilled water stored in air, the reaction was still capable of giving 66% yield (entry 10), whereas very little product was detected when the system was completely oxygen-free (entry 11). It should be noted that visible light does not alter the reaction efficiency.

Table 3

Studies on improving the compatibility of the reaction.

With the optimized reaction conditions in hand, a much more diverse series of aldehydes were selected to examine the substrate scope (Table 4). To our satisfaction, excellent yields were obtained with all aldehydes that we examined. Aromatic aldehydes where the –R group is a hydrocarbon (benzaldehyde, p-tolualdehyde, 5-indancarboxaldehyde, or 1-naphthaldehyde) were all transformed in quantitative or nearly quantitative yields (compounds 1 to 4). All of the electron-rich aromatic aldehydes that we tested—mono-, di-, and tri-methoxyl–substituted benzaldehydes—gave quantitative or nearly quantitative yields (compounds 6 to 9) regardless of the location of the substituent. Please note that piperonal, which was tested in our investigation of the reaction conditions (compound 5), also gave almost quantitative yield. The more hydrophobic 4-(pentyloxy)benzaldehyde and 4-(hexyloxy)benzaldehyde also gave excellent 94 and 90% yields (compounds 10 and 11), respectively. The 4-allyloxy-benzaldehyde gave quantitative yield as well, with the terminal C=C double bond intact and no observation of the Claisen rearrangement (compound 12), whereas the 4-benzyloxy-benzaldehyde resulted in a reduced 65% yield, probably due to the cleavage of the benzyloxy group (compound 13).

Table 4

Substrate scope investigation.

Other than those electron-rich aldehydes, only slightly reduced yields were obtained with 3-bromo-2,4-dimethoxybenzaldehyde and 5-bromo-1,3-benzodioxole-4-carboxaldehyde in which a bromine was also attached to the aromatic ring (compounds 14 and 15), possibly due to chelation. All other halogenated aromatic aldehydes, including fluorine-, chlorine-, bromine-, and the pseudohalogen cyano-substituted benzaldehydes, resulted in quantitative conversions regardless of the location of the substituent (compounds 16 to 24). With terephthalaldehyde, exclusive oxidation of only one of the aldehyde groups was obtained in a quantitative yield (compound 25). To verify this selectivity, introducing 4-formylbenzoic acid to our standard reaction conditions led to a quantitative recovery of the starting material. The presence of another carbonyl group, other than an acid, in the aldehyde does not affect the oxidation yield: both 4-acetylbenzaldehyde and 4-acetaminobenzaldehyde gave quantitative yields (compounds 26 and 27). 4-Hydroxymethyl-benzaldehyde also gave a quantitative yield (compound 28). With 4-quinolinecarboxaldehyde, a decreased yield (57%) was observed (compound 29), possibly due to the strong coordination of quinolone nitrogen. Other heterocyclic aromatic aldehydes such as furfural and 2-thiophenecarboxaldehyde were also subjected to the oxidation, in which furfural gave a quantitative yield, whereas 2-thiophenecarboxaldehyde gave a reduced 60% yield (compounds 30 and 31), possibly also due to the stronger coordination of the sulfur atom in thiophene. Most electron-poor aromatic aldehydes, represented by nitro- and trifluoromethyl-substituted benzaldehydes, were highly efficient in this oxidation and gave quantitative yields regardless of the location of the substituent (compounds 32 to 35).

Various aliphatic aldehydes were also tested: Linear hexanal, heptanal, octanal, and even the extremely hydrophobic decanal all gave quantitative oxidation products (compounds 36 to 39). Similarly, branched 2-methylbutanal, 2-methylpentanal, 2-ethylbutanal, and 2-ethyl hexanal also gave quantitative yields (compounds 40 to 43). With a C=C double bond being conjugated to the carbonyl, 3-methyl-2-butenal gave a slightly reduced yield of 77% (compound 44). Citronellal gave a moderate 60% yield, whereas citral gave a good 86% yield (compounds 45 and 46), indicating that the C=C double bond does interfere with the oxidation slightly. With aryl-substituted conjugated cinnamaldehyde and α-methylcinnamaldehyde, the oxidation proceeded quantitatively (compounds 47 and 48). Hydrocinnamaldehyde and phenylpropionaldehyde also gave quantitative yield (compounds 49 and 50). The 4-nitro–substituted cinnamaldehyde gave a slightly reduced 91% yield (compound 51). As a natural product and a widely used source of food spice, perillaldehyde also gave a quantitative yield (compound 52). An even more complex tertiary aldehyde, abietadien-18-al, was also successfully oxidized to abietic acid (compound 53) with an increased reaction temperature and pressure (Fig. 2).

Fig. 2

Abietadien-18-al aerobic oxidation.

Finally, a gram-scale oxidation was conducted with benzaldehyde (Fig. 3). To be practical and economical, we lowered both the amount of solvent and the amount of catalyst compared to the standard conditions. With only 2 mg of our silver(I) catalyst {equivalent to about 0.036 mol % [360 ppm (parts per million)] catalyst loading}, 560 mg of sodium hydroxide, and 1.4 ml of benzaldehyde in 10 ml of water at 1 bar of oxygen using an attached balloon, the reaction gave an astonishing 82% isolated yield with more than 1.4 g of analytically pure benzoic acid after 48 hours at 50°C. This indicates that our methodology can be readily scaled up to an industrial level.

Fig. 3

Gram-scale reaction.

On the basis of the results of our study, a plausible reaction mechanism that involves two catalytic cycles is proposed in Fig. 3: one of the cycles is responsible for extracting the hydride from the aldehyde, whereas the other is responsible for activating the dioxygen molecule in water. Each cycle consumes one molecule of aldehyde and generates one molecule of carboxylic acid. The two tandem cycles operate cooperatively to give the very efficient oxidation observed. It has previously been reported that the combination of silver(I) oxide and NHC ligand gives an NHC-Ag(I)-Cl complex (). After the chloride complex is introduced to our aqueous NaOH solution, the Cl− of the complex is substituted with hydroxyl to give the suggested NHC-Ag(I)-OH catalyst species. The catalyst then coordinates to the C=O double bond of the aldehyde and exchanges its coordinated –OH with the –H of the aldehyde, possibly through either a nucleophilic attack of –OH followed by β-hydride elimination (Fig. 4A) or a four-membered ring transition state where the exchange of –OH and –H occurred simultaneously (Fig. 4B). The catalyst then releases the carboxylic acid as the product and a silver(I)-hydride species, whose presence has been suggested by many of our recent studies (–) and has also been directly detected recently (). We also conducted a brief computational study for the proposed mechanism (detailed in the Supplementary Materials). To reduce the complexity of calculation, we used a simplified molecule model that simplifies the complicated NHC ligand into a simple trimethylphosphine ligand because of their electronic similarity. The result has shown that it is possible for the existence of a silver(I)-hydride intermediate when the silver center is coordinated to a strong electron-donating ligand (Fig. 4B). The computational result shows that a four-membered ring transition state is favored; however, considering the simplification of the calculation, there is still no solid proof of such a pathway. We tentatively propose that the silver(I)-hydride is responsible for the activation of oxygen in water, generating a silver(I)-hydroperoxyl intermediate. The hydroperoxide then nucleophilically attacks another carbonyl of the aldehyde. The hydrogen of the aldehyde is then extracted by silver with a similar β-hydride elimination (). Then, the silver(I)-hydride is oxidized by the coordinating peroxyl acid to give the corresponding carboxylate and to release a water molecule. The remaining carboxylate is then substituted by a hydroxide to release the carboxylate and regenerate the silver(I)-hydroxide catalyst.

Fig. 4

Reaction mechanism.

(A) Plausible mechanism and (B) ZPE-corrected energies from B3LYP/6-31G(d) and LANL2DZ for Ag. All values are given in kcal/mol and referred to the system AgOH(PMe3) + PhCHO. Gibbs free energy is considered for drawing.

In summary, we have discovered the first example of homogeneous silver(I)-catalyzed aerobic oxidation of an aldehyde in water. The reaction occurs at a very mild temperature, using water as the solvent and atmospheric oxygen as the oxidant, and this reaction can proceed with an extremely low loading of the Ag(I)catalyst (360 ppm catalyst on a gram-scale experiment). More than 50 different kinds of aldehydes were tested, and all underwent transformation to their corresponding carboxylic acids in mostly excellent to quantitative yields, indicating a good versatility and a variety of possible applications for this reaction. Further investigation of the mechanism and other potential applications of the silver(I) catalyst is currently under way in our laboratory.