Introduction

Imide functional groups have versatile applications in pharmaceutical, polymer, and natural product synthesis1,2. Traditionally imides have been synthesized by reaction between dicarboxylic acid e.g., phthalic acid (or anhydride) and amines3. Acylation of amides4, and rearrangement of isocyanates can also yield imides5. A few homogeneous catalytic oxidations of amides to imides are reported recently using organic hydroperoxides as terminal oxidants6,7,8.

Manganese oxides have gained wide recognition as powerful materials in catalytic oxidation reactions9,10. Readily accessible multiple oxidation states, high abundance, thermally stable structural forms, and dioxygen reduction ability are some of the important properties of these materials11,12,13. Several promoter ions can reside as structural or charge balancing ions to enhance the catalytic activity of manganese oxides14,15,16. Moreover, the diversity of synthetic methods allows preparation of different forms of manganese oxide of varying porosity and crystallinity such as octahedral molecular sieves (OMS), octahedral layer (OL), amorphous manganese oxide (AMO), birnessite, α-manganese oxide, and mesoporous manganese oxide11. These mixed valent, high surface area materials have excellent redox properties which lead to highly efficient manganese based oxidation catalysts. The effect of ion promotion in the manganese oxides has been studied previously17. One proposed mechanism is that with cesium ion promotion in manganese oxides, the surface tends to induce a more basic character, which allows the binding energy for lattice oxygen to be lowered17,18. Manganese oxide based materials have been utilized in different types of catalytic oxidation reactions such as selective oxidation of alcohols to aldehydes19, hydrocarbons to alcohols and ketones20, styrenes to styrene oxides21, alcohols to amides22,23,24,25,26, and amines to imines27,28,29. Furthermore, manganese oxide catalysts are effective in oxidative coupling of alkynes30, alkyne-silanes31, and alkyne-amine32. Herein, we report a heterogeneous catalytic oxidation of benzylic amides to imides by two different classes of manganese oxides.

Results and Discussion

N-benzylbenzamide Oxidation

Initially we utilized the oxidation of N-benzylbenzamide as the model reaction for developing the optimal reaction conditions33. In our early screening attempts, we used a number of redox active metal oxides (e.g., BiVO4, CuMoO4, CoMoO4) without much success. In a few cases, over-oxidation of the imide and thermal cracking reactions were observed. This was, however, not the case with Mn oxides. Building on our recently employed mesoporous manganese oxide material as a highly active oxidation catalyst17,34,35,36, a minute amount (0.14%) of electropositive Cs ions has been introduced in the manganese oxide structure, which enhanced the oxidation ability of manganese oxide by several orders of magnitude (100 fold) in oxidations of alcohols to aldehydes and amines to imines17,35. Herein, Cs+ promoted mesoporous manganese oxide (meso Cs/Mn2O3) was used as a model catalyst and acetonitrile as the model solvent. In early attempts, the use of air and oxygen as the oxidant did not produce any product (Entries 1 and 2, Table 1). Using H2O2 also did not give any product (Entry 3, Table 1). We then switched our attention to tert-butyl hydroperoxide (TBHP in water) as the oxidant. TBHP is a popular oxidant for oxidation of inert C-H bonds due to high thermal stability and solubility37,38 as compared to H2O2. The reaction was performed then with different loadings of catalyst and TBHP (Entries 1–4, Table S1). The reaction was then performed with solvents of different polarities at their respective boiling points (Entries 4–8, Table S1). No imide was detected using dioxane, THF, or toluene. Chloroform and CH2Cl2 were avoided due to potential generation of COCl2. (Entry 8, Table S1). Consequently, acetonitrile (ACN) was chosen as solvent for further reactions. The formation of another product (benzamide) can be attributed to the hydrolysis of intermediates.

Table 1 Optimization of oxidation of N-benzylbenzamide.
Full size table

tert-Butylhydroperoxide (TBHP) Decomposition

The rate of hydrolysis and subsequent formation of benzamide was increased with increasing concentration of the TBHP/water. To suppress the hydrolysis, TBHP in nonane (5.5 M, in molecular sieves) was used as the oxidant, minimizing the amount of water. Using TBHP in nonane, a maximum conversion of 10% and lower selectivity (15%) to imide (Entry 5, Table 1) was obtained. Negligible amounts of nonanol and nonanone were observed due to oxidation of nonane. To evaluate the rate of decomposition of TBHP, we probed the rate of decomposition of TBHP in the absence of amide substrate. Almost all of the TBHP was consumed within 45 min of reaction (Table S2). Alternatively, conjugated organic nitroxyl radicals combined with transition metal co-catalysts have been shown to be powerful catalytic systems for the auto-oxidation of hydrocarbons. Among those, N-hydroxyphthalimide (NHPI) is very popular in metal catalyzed oxidation of hydrocarbons39. Peroxyl radicals (ROO·) initiated by thermal decomposition of peroxides such as TBHP or deliberate introduction of free radical initiators, abstract an H-atom (>NO-H) from NHPI to form the PINO· radical (phthalimide N-oxyl)40. The combination of TBHP in nonane being added slowly (0.8 microliter/min) and using NHPI as an oxidation promoter (10 mol%) proved to be successful, as a significant increase in conversion (25%) was observed with high selectivity (90%) for imide (Entry 6, Table 1) was observed. Using molecular sieves as water scavengers, we surmised that adventitious water could be scavenged which would otherwise cause undesirable hydrolysis of imides. Thus the best conversion (90%) and selectivity (95%) towards imide were obtained (Entry 7, Table 1) with NHPI and molecular sieves.

Using air as oxidant decreased amide conversion (10%) significantly (Entry 8, Table 1) but showed the feasibility of using air as an oxidant. TBHP and NHPI without any catalyst did not produce any imide (Entry 12, Table 1); this signified the role of manganese oxide as catalyst in this reaction. Upon repeating the reaction under an argon atmosphere, we did not detect any imide, which proved the role of O2 as oxidant (Entry 9, Table 1). These results demonstrate the effectiveness of manganese oxides as heterogeneous catalysts for oxidation of amides to imides. The isolated yield (Entry 7) is in good agreement with the yield from GC-MS methods.

The oxidation of N-benzylbenzamide using Cs/Mn2O3 was then compared to different active, well known manganese oxide catalysts. Meso Mn2O3 having no Cs promoter ions showed much lower selectivity towards imide (Entries 1–2, Table 2). Cation vacancies in these oxides are the binding sites for amides based on our prior work where35 small loading (0.16% by weight) of Cs in the meso Cs/Mn2O3 material was critical for oxidation of amines to imines (which was used in this study). Using potassium containing manganese oxide octahedral molecular sieves (prepared by solvent free methods, K-OMS-2-SF) having similar Cs loading (0.16%), conversion and selectivity for imide decreased (Entry 4, Table 2). On the other hand, amorphous manganese oxide (AMO) prepared by redox methods displayed high conversion and selectivity (Entry 5, Table 2) to imides. Commercial manganese oxide was able to achieve minimal conversion without preferential selectivity towards the imide (Entry 6, Table 2). The reaction revealed no conversion in the absence of a catalyst (Entry 7, Table 2).

Table 2 Catalyst Scope for oxidation of N-benzylbenzamide.
Full size table

Substrate Scope

The substrate scope and limitations were then explored for different types of amides. This methodology works well for oxidation of activated (e.g, benzylic CH2) α-CH2 groups with respect to amide. AMO was able to oxidize aromatic (Entries 1, 2, 4–7, Table 3), lactam (Entry 6, Table 3), and heteroaromatic (Entry 7, Table 3) amides to corresponding imides with excellent conversion and selectivity. No side reaction was detected in the case of a benzamide with a para-chloro substitution (Entry 2, Table 3). However, to our surprise with ortho-chloro substitution in the benzamide, no imide formation was detected (Entry 3, Table 3). Oxidation of 1-isoindoline (Entry 6, Table 3) was facile and produced a cyclic imide with excellent conversion (>99%) and selectivity (100%). The yields from GC-MS studies are in reasonable agreement with isolated yields. A thiophene amide was converted to the imide effectively (Entry 7, Table 3) without oxidizing the sulfur to sulfoxide or sulfone. We used α methyl-substituted ortho carboxylic derivative of benzyl amide to investigate if a product related to the initial stage of oxidation could be isolated (complete oxidation to the imide would be impossible). However, the substrate was totally inert (Entry 8, Table 3), which may be due to difficult abstraction of N-H hydrogen atoms due to a steric effect of the methyl group. Unactivated aliphatic amides (Entries 9 and 10, Table 3) were not converted to imides using this procedure. This may be rationalized by the strength of α-CH bonds in those substrates that falls outside the range of the NHPI derived PINO radical’s C-H activation scope.

Table 3 Oxidation of amides to imides by manganese oxides.
Full size table

Kinetic and Stability Studies

Time dependent experiments for the oxidation of N-benzyl benzamide were conducted using meso Cs/Mn2O3 to study reaction rates (Fig. S1a). Periodic sampling was undertaken, the catalyst was separated by filtration, and conversion and selectivity of the filtrate were determined by GC-MS. Two separate time dependent experiments with and without NHPI were conducted. In the absence of NHPI, the reaction was very sluggish at the beginning (5% conversion after 8 h), whereas the presence of NHPI accelerated the reaction. Experiments with TBHP and NHPI indicated a first order rate equation with respect to amide (Fig. S1b) having a rate constant of 0.0025 min−1. PINO radical can abstract a hydrogen atom from amide (RCOCH2NH2) to initiate the reaction (Fig. S5). Since the addition rate of TBHP is slow, at the beginning the amount of TBHP was not enough for the generation of PINO radical. PINO radical reacts much faster than other peroxide based radicals with hydrocarbons, as confirmed by relative rate constants6. The catalyst stability was verified by performing reusability studies (Fig. S2b) with meso Cs/Mn2O3. After reaction, the catalyst was filtered, washed with excess solvent and water, dried under vacuum and reactivated at 250 °C for 30 min. No apparent loss of activity was observed after 3 catalytic cycles. Moreover, from the powder X-ray diffraction studies (Fig. S2a), it is apparent that the amorphous nature of the catalyst is retained after multiple reuses.

Possible Reaction Pathway

We investigated the relative rates of oxidation of para substituted N-benzylbenzamides (p-Cl, p-H, p-Me, p-OMe) to determine the electronic effect of the substituents. The Hammett equation can be used to interpret the electronic or steric influence of the substituents on the reaction intermediates. A linear relationship was found between ln (kX/kH) and the Brown–Okamato constant (σp+) (Fig. 1). The slope of the plot resulted in a reaction constant (ρ) value of 0.355, which signified involvement of a partial negative charge at the reaction center in the transition state of the rate-limiting step (polar effects in similar radical reactions are known)6.

Figure 1
figure 1

Hammett plot of competitive oxidation of para substituted N-benzylbenzamide. Reaction conditions are as described in Table 3 for 8 h. A linear relationship between ln(kx/kH) and Brown–Okamoto constant (σp+) for para substituted benzylamines with slope (ρ) of 0.355 was obtained, which indicates the formation of a negatively charged transition state.

Full size image

We propose a series of steps that can contribute to the formation of imides by manganese oxide (Fig. S3). Adsorbed amide molecules transfer an electron to the Mn center, followed by a cleavage of α C-H bond mediated by the NHPI/MnOx system (Fig. S3). The oxidation of an alpha substituted amide is almost undetectable under the present reaction conditions, where the presence of the methyl group could have hindered the binding of amide, thus blocking the cleavage of the α C-H bond. The reaction pathway in theory may involve an imine intermediate, by further removal of proton and electron transfer to Mn atoms (Fig. 2). However according to the DFT calculation performed at MO6-2X for this reaction in the gas phase, this N-H hydrogen atom abstraction will have a 12 kcal/mol higher activation energy (Ea) barrier than α C-H atom abstraction. The reduction of Mn centers simultaneously generates the formation of labile lattice oxygen, which reoxidizes the Mn center with production of H2O2, which can easily be decomposed over manganese oxide and form water1. Once formed, the α hydroperoxide intermediate can readily undergo a further abstraction of the 2nd benzylic hydrogen to yield the corresponding imide product. Hydroperoxide adducts were not detected by GC-MS, which signifies the rapid oxidation of the hydroperoxide species.

Figure 2
figure 2

The C-H abstractions vs N-H abstraction pathways.

Full size image

The aerobic oxidation pathway proceeds with meso Cs/Mn2O3 and AMO in the presence of NHPI. Perhaps a heterogeneous metal bound surface superoxide is formed as a first step41,42,43. This would then subsequently converted to a bound hydroperoxide followed by a similar mechanism TBHP as described in Fig. 2. This mechanism resembles the formylamide oxidation with peroxide and NHPI with a Co(II) catalyst. The exact mechanism is still under investigation with an emphasis on identification of all reactive oxygen species and potential correlation to reaction rates.

One of the major obstacles observed for many heterogeneous catalyst systems is of the product not desorbing from the catalyst surface. We wanted to study if there was a possibility that this phenomenon was occurring for the N-benzylbenzamide oxidation reaction; so, additional pre- and post-reaction characterizations were performed. X-ray Fluorescence spectroscopy (Table S3) was also done in order to further quantify the amount of product absorbed utilizing the most optimized conditions (Entry 1, Table 3). We were able to determine that there was approximately 5% by mass of the product still absorbed on the catalyst. These additional characterizations allowed us to partially account for the gaps between GC-MS and isolated yields.

Conclusion

In summary, manganese oxide materials can catalyze the oxidation of aromatic substituted amides to imides. Among the several materials studied, mesoporous Cs promoted manganese oxide (meso Cs/Mn2O3) and amorphous manganese oxide (AMO) were shown to be the best catalysts. AMO oxidized a diverse array of aromatic amide derivatives to produce imides with high conversions and selectivities (90–100%). Although aerobic oxidation with meso Cs/Mn2O3 mediated by NHPI gave modest yields of imide, in the optimal procedure TBHP was used as oxidant. This manganese oxide mediated selective partial oxidation method is useful as compared to precious metal catalyst (e.g., Pt, Rh, Pd or Ir) that are used for partial oxidation in terms of activity, cost, and environmental impact.

Methods

Synthesis of manganese oxide materials

The manganese oxide materials were synthesized by well-established methods meso Cs/Mn2O317, meso Mn2O317, OMS-2 (SF)44 and AMO45.

Synthesis of meso Cs/Mn2O3

Manganese (II) nitrate tetrahydrate (0.02 mol) was dissolved in a 1-butanol solution containing 0.188 mol (14 g) 1-butanol, 0.032 mol (2 g) of HNO3 and 3.4 × 10−4 mol (2 g) of P123 surfactant in a 150 mL beaker at RT and under magnetic stirring17. To this clear aqueous solution, 100 mL of 2.0 M CsNO3 was added maintaining the Mn:X ratio as 100:1. The resulting clear solution was then kept in an oven at 120 C for 3 h under air. The resulting black powder was washed with excess ethanol, centrifuged and dried in vacuum oven overnight. The dried black powder was then heated to 150 C for 12 h and 250 C for 3 h under air. The powder diffraction pattern of the material at this calcination temperature appears amorphous. However, upon heating at 450 °C or higher the XRD peak of this materials starts to appear crystalline with a pattern of Mn2O3 (Bixbyite)17. Meso Mn2O3was prepared using the same procedure without having the Cs ions.

Synthesis of OMS-2(SF)

In a typical experiment, 9.48 g (0.06 mol) of KMnO4 and 22.05 g (0.09 mol) of Mn(Ac)2‚4H2O powders were mixed and ground homogeneously in a mortar44. The mixed powders were then placed in an autoclave and heated at 80 °C for 4 h. The resulting black product was thoroughly washed with excess deionized water, and finally dried at 80 °C in air overnight.

Synthesis of AMO

A 60 mL volume of a 1.6 M Potassium permanganate solution was added dropwise to 100 mL of a 0.25 M Oxalic acid solution45. The mixture was stirred at room temperature for 2 hours. The resulting mixture was filtered and then washed with excess deionized water. The sample was then dried overnight at 90 °C.

General procedure for oxidation of amides to imides

In a typical amide oxidation reaction, a mixture of N-benzylbenzamide (0.5 mmol, 100 mg), catalyst (50 mg), N-hydroxyphthalimide (10 mol%), molecular sieves (4A) and acetonitrile (5 mL) was added in a 50 mL two necked round bottom flask equipped with a condenser. A solution of tert-butylhydroperoxide (TBHP, 10 equiv) in nonane (5.5 M) was added dropwise with a rate of 0.8 microliter min−1. The reaction mixture was heated to reflux under vigorous stirring (700 rpm) until the full consumption of TBHP. After reaction, the mixture was cooled and the catalyst was removed by filtration. The product analysis was done using GC-MS (gas chromatography-mass spectrometry). The conversion was determined based on the concentration of amides. Most reactions were repeated twice, and the average values were used. The imide products were isolated by silica gel column chromatography using hexane and (30–50%) ethyl acetate as the eluent and identified by 1H and 13C NMR spectra.

Characterization

The GC-MS analyses were performed with a 7820A GC system connected with a mass detector of 5975 series MSD from Agilent Technologies and a nonpolar cross-linked methyl siloxane column with dimensions of 12 in × 0.200 mm × 0.33 µm was used. The 1H and 13C NMR spectra were recorded on a Bruker AVANCE III- 400 MHz spectrometer. 1H NMR spectra were collected at 400 MHz with chemical shift referenced to the residual CHCl3 peak in CDCl3 (δ: H 7.26 ppm). 13C NMR spectra were collected at 100 MHz and referenced to the CDCl3 signal (δ: C 77.0 ppm)46. Only in case of phthalimide the solvent was DMSO-d6, and chemical shifts were referenced to the residual DMSO-d5 peak in DMSO-d6 (δ: H 2.50 ppm) for 1H NMR and the DMSO-d6 peak (δ: C 39.51 ppm) for 13C NMR46. The spectral data of imide products were compared with the literature reports47.