Samantha Caputo‡
a,
Alessandro Kovtun‡b,
Francesco Brunocde,
Enrico Raveracdef,
Chiara Lambruschinia,
Manuela Meluccib and
Lisa Moni*a
aDepartment of Chemistry and Industrial Chemistry, University of Genoa, Via Dodecaneso 31, 16146 GENOVA, Italy. E-mail: lisa.moni@unige.it
bConsiglio Nazionale delle Ricerche, Istituto per la Sintesi Organica e la Fotoreattività (CNR-ISOF), Via Gobetti 101, 40129 BOLOGNA, Italy
cMagnetic Resonance Center (CERM), University of Florence, Via L. Sacconi 6, 50019 Sesto Fiorentino, Italy
dDepartment of Chemistry “Ugo Schiff”, University of Florence, Via della Lastruccia 3, 50019 Sesto Fiorentino, Italy
eConsorzio Interuniversitario Risonanze Magnetiche di Metalloproteine (CIRMMP), Via L. Sacconi 6, 50019 Sesto Fiorentino, Italy
fFlorence Data -scienze, University of Florence, Italy
First published on 26th May 2022
The carbocatalyzed synthesis of 2,3-disubstituted quinolines is disclosed. This process involved a three-component Povarov reaction of anilines, aldehydes and electron-enriched enol ethers, which gave the substrate for the subsequent oxidation. Graphene oxide (GO) was exploited as a heterogeneous, metal-free and sustainable catalyst for both transformations. The multicomponent reaction proceeded under simple and mild reaction conditions, exhibited good functional group tolerance, and could be easily scaled up to the gram level. A selection of tetrahydroquinolines obtained was subsequently aromatized to quinolines. The multistep synthesis could also be performed as a one-pot procedure. Investigation of the real active sites of GO was carried out by performing control experiments and a by full characterization of the carbon material by X-ray photoelectron spectroscopy (XPS) and solid-state nuclear magnetic resonance (ssNMR).
Recently carbon nanomaterials, with particular reference to graphene oxide (GO), have shown to exhibit interesting catalytic activity in organic transformations,16–18 such as C–C bond formation,19,20 C–N and C–O oxidation,21–23 thiol oxidation24 and different MCRs.25–32
GO is a π-conjugated planar material characterized by the following structure and active sites:16,18,33 a few nm large aromatic Csp2-domains surrounded by oxygen functional groups (e.g. alcohols, epoxides and carboxylic acids) as well as free radicals, which are due to internal structural defects. Because of this multifunctional character, GO can be considered as a multicatalytic system, intrinsically versed towards one-pot multistep syntheses. Nevertheless, GO-assisted one-pot multistep reactions have been seldom explored,34 at least using organo- or metal-functionalized GO composites.35 Building on our experience in the development of new synthetic methodologies using MCRs36–44 and in one-pot multistep processes,45,46 and inspired by the potential of MCRs involving the employment of an external oxidant,47 we decided to study the GO-promoted Povarov multicomponent reaction between anilines, aldehydes and 2,3-dihydrofuran (DHF) that acts as dienophile to give tetrahydroquinolines (THQs) 1, and their subsequent oxidation to 2,3 disubstituted quinolines 2 (Scheme 1).
Scheme 1 Syntheses of 2,3-disubstituted quinolines 2 via the Povarov multicomponent reaction and subsequent oxidation. |
In this context, the dual capacity of GO to serve both as an acid and an oxidant should favour the in situ formation of imine, its activation in the aza-Diels–Alder process and the subsequent oxidation of the THQ adduct into the corresponding quinoline.
THQs48 and quinolines49,50 are privileged heteroaromatic scaffolds, largely present in the structures of bioactive compounds and natural products, with many applications in different chemical domains, such as organic synthesis or material science.
Even though the Povarov reaction has been extensively investigated for the synthesis of THQs, some drawbacks persist: strongly acidic conditions, use of noxious and expensive catalysts, formation of mixtures of products, unsatisfactory yields, limits on the scope (i.e. aliphatic aldehydes), and longer reaction times.51
In the literature, the routes to 2,3-disubstituted quinolines via a Povarov multicomponent reaction and subsequent oxidation are still relatively underexplored, presenting either a limited scope52,53 or requiring multistep procedure (Scheme 1).54 In 2018, Meléndez Gómez reported a simple and efficient one-pot synthesis of quinolines 2; however the scope in that case was restricted to 2,3-di(hydroxyalkyl)quinolines.55
Herein, we report the first GO-promoted one-pot multistep synthesis of substituted quinolines and the investigation of the real active sites of GO using X-ray photoelectron spectroscopy (XPS) and solid-state nuclear magnetic resonance (ssNMR).
After the reaction mixture was stirred at room temperature for 24 h, the furo-[3,2-c]tetrahydroquinoline 1a was isolated in a 56% yield as a mixture of endo- and exo-isomers (Table 1, entry 1).
Entry | Solvent (M) | Dienophile (equiv.) | Temperature | GO (mg mmol−1) | Time (h) | endo/exo ratioa | Yield of 1ab |
---|---|---|---|---|---|---|---|
a endo/exo ratio was calculated after purification by column chromatography.b Isolated yield.c The reaction was performed with 1.5 equiv. of the dienophile; after 2 h, an additional amount of 1.5 equiv. was added and the mixture was stirred further for 2 h.d Reaction was performed in the presence of MnCl2 4H2O (0.12 mg mmol−1): the amount of the catalyst was decided based on the amount of Mn contained on GO (about 5000 ppm, see ESI for details).e reaction was performed in the presence of reduced graphene oxide (rGO). | |||||||
1 | CH3CN (0.4) | 1.5 | rt | 20 | 24 | 80:20 | 56% |
2 | CH2Cl2 (0.4) | 1.5 | rt | 20 | 24 | — | — |
3 | Toluene (0.4) | 1.5 | rt | 20 | 24 | — | Trace |
4 | Pentane (0.4) | 1.5 | rt | 20 | 24 | — | 18% |
5 | EtOAc (0.4) | 1.5 | rt | 20 | 24 | 80:20 | 22% |
6 | 1,4-Dioxane (0.4) | 1.5 | rt | 20 | 24 | 80:20 | 13% |
7 | 2-Me-THF (0.4) | 1.5 | rt | 20 | 24 | — | 20% |
8 | DMF (0.4) | 1.5 | rt | 20 | 48 | 60:40 | 36% |
9 | iPrOH (0.4) | 1.5 | rt | 20 | 24 | 90:10 | 43% |
10 | tBuOH (0.4) | 1.5 | rt | 20 | 24 | 70:30 | 48% |
11 | H2O (0.4) | 1.5 | rt | 20 | 24 | — | 20% |
12 | CH3CN (0.4) | 1.5 | 60 °C | 20 | 24 | 80:20 | 27% |
13 | CH3CN (0.4) | 1.5 | 10 °C | 20 | 24 | — | — |
14 | CH3CN (0.4) | 1.5 | rt | 40 | 4 | 80:20 | 57% |
15 | CH3CN (0.8) | 1.5 | rt | 20 | 4 | 80:20 | 60% |
16 | CH3CN (0.8) | 1.5 | rt | 10 | 4 | 80:20 | 42% |
17 | CH3CN (0.8) | 3 | rt | 20 | 4 | 75:25 | 62% |
18 | CH3CN (0.8) | 1.5 + 1.5c | rt | 20 | 4 | 8020 | 62% |
19 | CH3CN (0.8) | 1.5 | rt | — | 4 | — | <5% |
20d | CH3CN (0.8) | 1.5 | rt | — | 4 | 80:20 | 20% |
21e | CH3CN (0.2) | 1.5 | rt | 20 (rGO) | 6 | 75:25 | 21% |
Starting from this result, different reaction conditions were explored (Table 1). The effect of different solvents was first evaluated. Apolar solvents, such as pentane, toluene, dichloromethane (entries 2–5), or ethers (entries 6,7) afforded very low conversions and poor yields. Better results were obtained with protic solvents, such as iPrOH (entry 9) and tBuOH (entry 10), although the concomitant formation of the corresponding acetal decreased the yield (the corresponding di-isopropyl acetal was isolated in a 25% yield using iPrOH). In DMF (entry 8), the reaction turned out to be very slow: based on the literature data,56 we suppose a role of DMF in stabilizing the imine and deactivating the GO surface by covalent bonding. Having assessed that the best results were obtained by using CH3CN as solvent, we investigated other reaction parameters, such as the temperature (entries 12 and 13), concentration (entry 15), and amount of catalyst and dienophile (entries 14–18). We found that the best conditions in terms of yield, atom economy and environmental impact, were the following: CH3CN 0.8 mol dm−3, in the presence of 20 mg mmol−1 of GO, 1.5 equiv. of DHF for 4 h at room temperature. The stereochemistry of the products was determined by the combined use of the coupling constant and literature data.57
Proofs of a genuine GO catalysis were gained via dedicated control experiments. The background reaction was not productive (entry 19), while the potential co-catalysis promoted by the metal contamination of GO (i.e. manganese(II)) was ruled out via a designed experiment with MnCl2 4H2O (entry 20).58 Using reduced GO (rGO) as a carbocatalyst, the desired product was obtained in a poor yield, proving the important role of the oxygenated groups (entry 21).
In order to evaluate the versatility of the synthetic protocol, we moved on to establish the scope, with varying the aldehyde, aniline and dienophile, under the optimized reaction conditions (Scheme 2).
Scheme 2 Scope of the carbocatalyzed Povarov reaction. Different reaction conditions: (a) rt, 24–48 h; (b) 60 °C, 24–48 h; (c) CH3CN/H2O = 4:1, 60 °C, 24–48 h. |
Good results were obtained when performing the reaction with DHF as dienophile: interestingly, good yields were observed employing aliphatic (1j and 1k) and α,β-unsaturated (1l) aldehydes, substrates rarely used in the Povarov reaction;54,59,60 on the other hand, substituted anilines (1b, 1f and 1p) and 2,3-dihydropyran (DHP) as dienophile (1q and 1r) gave lower yields. To improve the conversion of the less reactive substrates, we slightly changed the reaction conditions, applying longer time and/or higher reaction temperature, affording the products in acceptable yields and with endo as the major isomer. For compound 1r, where surprisingly the inversion of diastereoselectivity was observed, we tried to improve the yield by using a mixture of CH3CN/H2O as 4:1,61 showing an increase in the yield from 28% to 40%.
Different behaviour was observed by using cyclopentadiene as dienophile: even though the reaction a required longer time (24 h) and higher temperature (60 °C), due to the lower reactivity of the dienophile (4 equiv.), the corresponding THQ 1s was obtained with a 70% yield and high diastereoselectivity. Ethyl vinyl ether (1m–o) gave good yields, even though it required a longer reaction time.
Reaction with benzaldehyde, 3,5-dimethoxyaniline and DHF afforded a complex mixture of different products and the corresponding tetrahydroquinoline 1u was not observed.
Employing benzofuran as dienophile, no reaction occurred, despite heating at 60 °C for 24 h to promote the reactivity.
To assess the efficiency of the catalyst and verify its recyclability, we performed a scale-up experiment using 7.5 mmol of p-chlorobenzaldehyde (Scheme 3). In this case, a notable 66% isolated yield of 1a was obtained under slightly different conditions (CH3CN 0.2 mol dm−3, 2.5 equiv. of DHF, 24 h).
At the end of the experiment, the solvent was removed by evaporation and the product was recovered by dissolving the mixture in dichloromethane/ethyl acetate followed by centrifugation at 7800 rpm for 5 min. The catalyst was washed by repeating the process 10 times to remove all traces of organic compounds. The recovered catalyst was washed again with water and methanol, and subsequently reused in consecutive reactions. A slight deactivation of the GO catalyst was noticed after the first run, while a significant decrease was noted from the 4th run. The above results indicate that the GO catalyst showed moderate recyclability within three runs (Fig. 1).
After the recyclability experiments, the recovered GO after the first and the last run were analyzed by XPS and ssNMR. XPS showed a significant decrease in the epoxide and hydroxide groups during the reaction (Fig. 2). The recyclability experiments were repeated with other batches of GO (one produced by Graphenea, and one by Abalonyx, which contained a much lower amount of Mn), giving similar results in terms of the yields and XPS analyses (see ESI for details†). The XPS survey spectra of GO in the reaction conditions (control sample of GO in CH3CN at room temperature for 24 h) and GO after multiple reactions (see ESI for details†) were mainly composed of O 1s and C 1s signals; the presence of N 1s, Si 2p, S 2p, Mn 2p and Cl 2p signals were found at lower intensities. The binding energies and relative atomic concentrations of all the samples are reported in the ESI†.
Fig. 2 XPS C 1s signal of: (a) commercial Graphenea GO; (b) GO after CH3CN at room temperature for 24 h; (c) GO after the 1st cycle; (d) GO after the 6th cycle. |
The XPS C 1s signal analysis showed a significant decrease in C–O groups content and a concomitant increase in the CC Sp2 and C–C Sp3 regions. However, the efficiency of the reaction might not be directly related to the overall oxidation degree of the material, but rather to the increase in the amounts of nitrogen (N) and chlorine (Cl).
In particular, N was so far present in both GO Abalonyx and GO Graphenea, about 0.3–0.4% in the control samples. After the first and last reactions, the amount of N proportionally grew, and the efficiency seemed to decrease as a function of N (Fig. 3a). The Cl showed a less evident behaviour: the initial amount in GO Graphenea was negligible, but in GO Abalonyx, it was initially present in a significant amount (0.5%). If we consider only the trend in GO Graphenea, the amount of Cl grew after the first and the last reactions (Fig. 3b). Since the increase in the amount of nitrogen was higher than the increase measured for chlorine, the presence of these two elements seemed to be associated with the interaction of the aldehyde and aniline with the GO surface, rather than with the final product.
The simplest interpretation of the N and Cl trends is that the p-chlorobenzaldehyde and aniline are bound to the GO surface, blocking the active sites and preventing further Povarov reaction. Obviously, the presence of these molecules on GO affected the O/C decrease measured.
In order to clarify the nature of the interaction between the organic molecules and GO, we treated the recovered GO under acidic conditions,62 and we repeated the XPS analysis, finding no significant variation in GO's chemical structure: the O/C ratio slightly decreased, but N and Cl amounts were identical to in the untreated sample. Acid treatment apparently was not able to regenerate the GO as previously reported,61 probably due to the presence of grafted reagents on the active sites. Moreover, we carried out the Povarov reaction using the GO recovered after acidic treatment, but it gave 1a in only a 17% yield.
To investigate further the GO functionalization after the Povarov reaction, characterization by NMR spectroscopy was performed. The 13C ssNMR spectra showed peaks corresponding to aromatic C-sp2 at 130 ppm, hydroxyl groups at 70 ppm, and epoxy at 60 ppm; carboxyl was present in lower amounts.63 From the quantitative 13C ssNMR spectra obtained by direct excitation (see Fig. S7 and S8 in the ESI†), a decrease in C–OH and C–O–C groups was observed; in fact, the epoxy ring opening is one of the most common mechanism for covalent grafting, as reported in the literature through XPS and NMR,64 but in the present study it could be excluded: the C–O–C/C–OH ratio obtained by ssNMR was close to 2 for all the samples. The overall reduction of GO after the Povarov reaction, observed by XPS, could be confirmed also by ssNMR. The signal for the aliphatic carbon (Csp3) appeared at 31 ppm in both the direct excitation and cross-polarization spectra of GO after the reaction (Fig. 4).
Fig. 4 CP 1H–13C ssNMR signals of commercial Abalonyx GO after CH3CN at room temperature for 24 h; GO after the 1st cycle; GO after the 3rd cycle. |
The 1H–13C wPMLG-HETCOR spectra65–67 (Fig. 5) presented a correlation between the signal of 13C at 31 ppm and a signal in the 1–2 ppm region in 1H, further proof of the formation of aliphatic carbons on GO surfaces. In the literature, there are a few reports of examples of saturated C–H bond signals in the 30 ppm region, such as the NaOH-promoted Suzuki cross-coupling reaction of phenylboronic acid68 or the methylene groups in the covalent grafting of amines.64 The relative amount of (Csp3) obtained from the direct excitation 13C NMR spectra (14% ca) seemed to be independent from the GO type and number of cycles.
Fig. 5 1H–13C correlation ssNMR signals of commercial Abalonyx GO after CH3CN at room temperature for 24 h (a); GO after the 1st cycle (b); GO after the 3rd cycle (c). |
Summarising, XPS and NMR analysis showed that: i) GO after the Povarov reaction was significantly reduced, (ii) no epoxy ring opening was observed, and iii) p-chlorobenzaldehyde and aniline could be covalently grafted, as well as dihydrofuran, while iv) the presence of aliphatic Csp3 may suggest that a fraction of reaction sites was inactivated after the reaction.
Although the mechanism of the Povarov reaction has not yet been certainly established, a step-ways manner is generally proposed rather than a concerted one:69,70 The in situ-formed imine, activated by BrØnsted acid (BA) or Lewis acid (LA) catalysts, undergoes nucleophilic attack from the dienophile, with the formation of a cationic intermediate, which evolves into the final THQ after a quick intramolecular Friedel–Crafts reaction. Actually, GO presents both BA and LA properties. Indeed, while the BA character of GO has been well established, recently, the Lewis acidity, due to the basal plane epoxide ring opening, was found to be crucial for its catalytic activity involving multicomponent reactions.62 Therefore, based on our experimental findings and on the previous literature, a possible mechanism is proposed and depicted in Scheme 4. Initially, the BA groups on GO promote the protonation of the aldehyde and, after nucleophilic attack by aniline, the dehydration step, give imine I. GO can donate a proton to the imine I, forming the reactive iminium species II, which undergoes reaction with DHF, following the generally accepted mechanism of the Povarov reaction, to generate 1a. Another possibility is that GO might act as a Lewis acidic carbenium catalyst,71 interacting with imine to give complex III; the electron-rich olefin, such as DHF, attacks complex III to form intermediate IV, which undergoes an intramolecular cyclization to produce complex V. Then, the detachment of the product with subsequent aromatization provides THQ 1a.
Based on our experimental results, the decrease in efficiency was related to the partial passivation of the surface of the material due to the covalent bonds with organic substrates. Although a direct covalent interaction between GO and the Povarov product cannot be excluded (for instance, the covalent bond with intermediate V), the reagents may interact with the surface independently. Actually, aniline is able to chemically interact with the GO surface, opening an epoxy group and/or reacting with the Cα of a hydroxyl group.72 Eventually, alcohols or amino-alcohol moieties grafted on GO should interact with the aldehyde in an unproductive pathway with partial deactivation of the carbon material. Moreover, the appearance of new peaks at 31 ppm in the 13C-NMR ssNMR spectra suggested the interaction of DHF as well. This molecule can undergo GO-promoted acid-hydration, to afford the corresponding hemiacetal, which can covalently interact with the GO surface, increasing the passivation.
Then, we proceeded to investigate the carbocatalized oxidation of the Povarov products to give substituted quinolines. GO has been largely exploited for the oxidation of organic compounds,22,23,73,74 although very few examples have appeared recently in the literature regarding the oxidation of heterocycles.75
Aiming to develop a one-pot procedure, we initially treated THQ 1a (as a diastereoisomeric mixture) in conditions similar to those used for the Povarov reaction, except for the temperature, which was increased to 110 °C. Contrary to what is reported in the literature with the most commonly used oxidant 2,3-dichloro-5,6-dicyano-1,4-benzoquinoneone (DDQ),54 where a mixture of different products is generally obtained, the reaction provided the 2,3-disubstituted quinoline 2a as a single product, even though it proceeded very slowly, affording a low yield after 4 days (Y = 28%) (Table 2, entry 1). Product 2a was the result of an elimination–oxidation process, generally promoted by acidic conditions.76,77 In addition, in our case, based on the ratio of endo/exo isomers of the starting material recovered, only the endo isomer appeared to have reacted, while the exo isomer remained untouched.
Entry | Catalyst (mg mmol−1) | Solvent (M) | Time (h) | Temperature | Molar ratio (1a/2a)a | Yield exo 1a | Yield 2ab |
---|---|---|---|---|---|---|---|
a Conversion was calculated after purification by column chromatography.b Isolated yield.c Reaction was performed starting from a mixture endo/exo of 80:20.d Reaction performed starting from pure endo.e Reaction was performed in the presence of Na2CO3 (40 mg mmol−1). | |||||||
1c | GO (20) | CH3CN (0.2) | 96 | 110 °C | 50:50 | — | 28% |
2d | GO (50) | CH3CN (0.2) | 48 | 120 °C | 35:65 | — | 56% |
3d,e | GO (50) | CH3CN (0.2) | 48 | 120 °C | 55:45 | — | 44% |
4d | GO (50) | CH3CN (0.1) | 96 | 120 °C | 20:80 | — | 49% |
5d | GO (50) | CH3CN (0.1) | 96 | 90 °C | 40:60 | — | 45% |
6d | GO (50) | CH3CN/H2O 4:1 (0.2) | 48 | 120 °C | 15:85 | — | 63% |
7d | GO (100) | CH3CN (0.1) | 48 | 120 °C | 5:95 | — | 88% |
8d | GO (100) | CH3CN/H2O 4:1 (0.1) | 24 | 120 °C | 0:100 | — | 92% |
9d | GO (100) | CH3CN/H2O 4:1 (0.1) | 4 | 120 °C (MW) | 20:80 | — | 71% |
10d | GO (100) | CH3CN/H2O 2:1 (0.1) | 3 | 120 °C (MW) | 10:90 | — | 88% |
11c | GO (100) | CH3CN/H2O 4:1 (0.1) | 24 | 120 °C | 15:85 | 13% | 73% |
To simplify the optimization, at first, we decided to study the oxidation reaction starting from the pure endo (easily accessible by trituration of the mixture). Then, the reaction was carried out with a slight increase in the temperature and catalyst, leading to a better, albeit still unsatisfactory, conversion and yield (entry 2). Based on the work of Zang,75 we repeated the oxidation in the presence of a substoichiometric amount of Na2CO3, suggesting that the base does not have a role in the process (entry 3). Lower temperatures or longer times did not produce significant improvements, while the use of a mixture of CH3CN/H2O as 4:1 afforded a great conversion of 1a endo into quinoline 2a (Y = 63%, entry 6). Furthermore, the increase in GO to 100 mg mmol−1 (entry 7) was explored, while performing the reaction in CH3CN 0.1 M to allow a better dispersion of the catalyst in the solvent, giving almost a complete oxidation of 1a endo (Y = 88%). The best reaction conditions were found to be the use of a mixture of CH3CN/H2O as 4:1 (0.1 M) and 100 mg mmol−1 of GO; indeed, complete conversion of THQ endo into quinoline was observed, affording the desired adduct in a 92% isolated yield (entry 8).
To shorten the reaction time, we tried to perform the oxidation process under microwave (MW) irradiation (entries 9 and 10). This provided an outstanding conversion and time reduction, and quinoline was obtained in an 88% isolated yield.
Finally, we tried the oxidation of a diastereoisomeric mixture of 1a (endo/exo = 80:20) under the optimal reaction conditions, obtaining a complete conversion of 1a endo into quinoline 2a (73%) and partial recovery of 1a exo (13%) (entry 11).
To explain the mechanism of the oxidation and its diastereoselectivity, we performed several experiments (Scheme 5). Since the oxidation–elimination product is generally formed under acidic conditions by elimination and subsequent aromatization by the air, we were interested to verify if GO was only effective as an acid promoter just in the first step (Scheme 5a, pathway A). With this aim, N-substituted THQ 3 was synthetized through a reductive amination reaction and reacted under the optimized conditions, recovering the unaltered starting material (Scheme 5b). This result suggests that elimination might not be the first step, or, in any case, the GO does not only mediate the elimination step. However, could GO catalyze first the C–N oxidation, and the intermediate then be aromatized by the elimination process (Scheme 5a, pathway B)? To investigate this possibility, THQ 1s, where the elimination is prevented by the lack of oxygen, was treated under the optimized conditions, but also in this case only the starting material was recovered (Scheme 5c).
Based on these experimental results, we suggest that the process is concerted and that the interaction between THQ and GO surface is expected to be more favourable for the endo isomer where both hydrogens have a cis relationship, so the steric hindrance of the aryl group does not prevent the simultaneous interaction between the THQ core and the oxygen of the tetrahydrofuran moiety.
Moreover, other control experiments were performed to investigate the catalytic activity and the role of GO and oxygen in the process (Table 3). The potential co-catalysis promoted by metal contaminations of GO was ruled out by performing the reaction with MnCl2 4H2O (entry 2): actually, no reaction occurred; as well as by heating THQ 1a without GO as a control experiment (entry 1).
Entry | Variation from standard conditions | Yield |
---|---|---|
a MnCl2 4H2O was used instead of GO considering that GO contains 5000 ppm of Mn by ICP-OES analysis, the control experiment was performed by using the equivalent amount of Mn(II) (see ESI for details). | ||
1 | No GO | 11% |
2 | MnCl2·4H2O (0.12 mg mmol−1)a | — |
3 | GO (100 mg mmol−1) under Ar | 53% |
4 | rGO (100 mg mmol−1) | 53% |
5 | rGO (100 mg mmol−1) under Ar | 21% |
The carbocatalytic oxidation carried out under an argon atmosphere (entry 3) brought a low conversion with a 53% isolated yield, demonstrating that oxygen is important, although its absence does not completely suppress the reaction. Furthermore, when reduced graphene oxide (rGO) was employed, the same decrease in yield was noted (entry 4).
These results suggested that both oxygen and hydroxyl and epoxy groups on GO play a role during oxidation. As a further demonstration of this, the yield dramatically decreased when the reaction was carried out with the concomitant presence of rGO and an argon atmosphere (entry 5). Based on the experimental evidence and literature data,21 we deduce that GO serves both as an oxidant, thanks to the hydroxyl and epoxy groups, and as a catalyst in the aerobic oxidation of the substrate. Further proof of this dual mechanism was obtained from the recyclability experiments. The reusability of GO was investigated following the same protocol used for the MCR. Even though the XPS analysis disclosed a significant reduction in the carbon material after 3 runs, the yield dropped only up to 50%, similar to what happened when using rGO (see ESI for details†). The magnitude of the GO reduction was significantly higher than for GO after the Povarov reaction: after the first oxidation cycle the O/C dropped from 0.28 ± 0.02 for the control sample to 0.12 ± 0.01, as well as all the C–O groups decreased, as revealed by C 1s XPS analysis: the epoxide and hydroxide contracted from 22.4% and 6.6% to 5.2% and 3.1%, respectively.
Finally, an oxidation using GO recovered from the 3rd run of the Povarov reaction was performed, obtaining 2a in a high yield (81%). This result suggests that the two processes might involve different functional groups on the GO surface, making the one-pot process feasible.
Next, the scope of the optimized protocol was investigated for a variety of tetrahydroquinolines (Scheme 6). The reaction proceeded smoothly, and complete consumption of the endo isomer was observed almost in each case. Quinolines bearing different aromatic substituents at position 2 were obtained in excellent yields. Interesting results were obtained even with compounds bearing alkyl or unsaturated substituents, allowing the isolation of products 2j, 2k and 2l in good to excellent yields.
Unfortunately, THQ 1o gave the corresponding quinoline 2o in a poor yield due to the extended degradation.
Finally, having demonstrated the efficiency of this protocol, we investigated the one-pot procedure leading to 2, without isolating the Povarov product 1 (Table 4). After a short optimization (see ESI†), we found the best results employing an increased dilution in the Povarov step. Then, as soon as the multicomponent reaction was complete, acetonitrile/water solution and fresh GO were directly added and the mixture was heated at 120 °C for 48 h.
Although a complete conversion of THQs has never been observed, the final quinolines were isolated in good yields, if compared with the previously reported two-step procedure. In two cases (Table 4, entries 1 and 3), the one-pot sequence was superior; whereas for compounds 2h and 2j (entries 2 and 4), the two-step sequence seemed to be more efficient. Interestingly, we were also able to isolate the not-reacted THQ as a mixture of diastereoisomer with exo as the major one, which confirmed its lower reactivity in the oxidation step.
In addition, we also wanted to demonstrate that the presence of an additional functional group on the aromatic ring (2i) may be exploited for further derivatizations. As polycycle-fused quinolines are typical privileged structures, widely distributed in natural products and medicinal chemistry,78–81 we studied the conversion of compound 2i into the benzo-oxepin-fused quinoline 4 by hydrogenolysis of the benzyl protecting group, followed by a Mitsunobu reaction with tbutyl azodicarboxylate (TBAD) and triphenylphosphine.82
The desired product was isolated after column chromatography in excellent yield (Scheme 7). Interestingly, this simple protocol allowed the synthesis of one of the rare examples of an oxepine-containing polycyclic-fused quinoline system.83–86
XPS spectra were acquired by a hemispherical analyser (Phoibos 100, Specs, Germany) using a non-monochromatic Mg Kα excitation set at 125 W (XR50, Specs, Germany). Survey and high resolution spectra was acquired in fixed analyser transmission mode (FAT) on a large area of ca 7 × 3 mm2 and overall energy resolution of 0.9 eV measured on freshly sputtered silver (Ag 3d). The spectrometer was calibrated to the Au 4f7/2 peak at 84.0 eV. Static charging effects were corrected for by calibrating all the spectra to C 1s 285.0 eV. Deconvolutions were performed using CasaXPS software after Shirley background subtraction. C 1s was fitted as described in the literature,87 using the asymmetric line-shape for aromatic C-sp2 and symmetric line-shapes (pseudo-voigt) for the C–O defects. The binding energies of the synthetic components of C 1s were: CC sp2 at 284.4 eV, CC* sp2 at 283.6 eV, C–C sp3 285.0 eV, C–OH at 286.2 eV, C–O–C at 286.9 eV, CO at 288.2 eV and O–CO at 289.1 eV. The O/C ratio was obtained from the O 1s and C 1s area ration and from C 1s fitting according to the stoichiometric ratios of C–O groups.
Solid-state NMR experiments were recorded on a Bruker Avance II spectrometer operating at 700 MHz 1H Larmor frequency (16.4 T), corresponding to 176 MHz 13C Larmor frequency. The spectrometer was equipped with a 3.2 mm BVT MAS probehead in double resonance mode. The magic angle spinning (MAS) frequency of the sample was set to 11.111 kHz. In the 1D direct excitation 13C-NMR spectra, the excitation pulse duration was set to 3.0 μs, corresponding to a 56° flip angle, the spectral window was 625 ppm, and the interscan delays were optimized to 1.9 s to make the experiment quantitative. For the 2D {1H}-13C HETCOR experiments, cross-polarization was achieved by matching the k = 1 Hartmann–Hahn condition. The 90° pulse duration on 13C and 1H were set to 4.8 us and 3.3 us, respectively. The spectral windows for 1H and 13C were 20 and 250 ppm, respectively. During the 1H magnetization evolution under the chemical shift in the indirect dimension, the PMLG decoupling sequence was used to suppress 1H–1H dipolar couplings. In these experiments, the interscan delay was set to 1 s.
The {1H}-13C HETCOR spectra were coprocessed for denoising as recently proposed by some of us,88 and processed applying a square-sine window function on both time dimensions. The presence of paramagnetic impurities originating from Mn ions is particularly critical for the spectrum acquisition of pristine GO produced by Graphenea as they present a relatively high Mn residual, in accordance with the results published by Panich.89 Hummers derived commercial GOs (Abalonyx and Graphenea) were produced by using also permanganate and the cleaning procedures may leave a small fraction of Mn: 0.15 ± 0.03 at% by XPS and 0.5% by ICP-OES in Graphenea and below 0.03%, the sensitivity limit for XPS, in Abalonyx. GO Abalonyx presents less experimental problems and, luckily, the control sample of GO Graphenea presented a significantly lower amount of Mn with respect to the pristine sample. On the other side, the reduced graphene oxide, rGO, presented also paramagnetic properties due to the high C-sp2 content (above 70% of C atoms).
Footnotes |
† Electronic supplementary information (ESI) available. See https://doi.org/10.1039/d2ra01752k |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2022 |