Facile synthesis of novel heterogenous Rh/COF catalyst and its application in tandem selective transfer hydrogenation and mono-methylation of Nitro compounds with Methanol

The development of transition metal heterogeneous catalysts for economical and effective synthesis of N-methylamine, especially for the mono-methylation of amines is still challenging. Herein, two unprecedented Rh-supported COFs heterogeneous catalysts Rh/MelCOF was facile synthesized by Schiff base reaction using melamine as a precursor, and for the first time, it was successfully applied to the effective and high selective tandem reaction of transfer hydrogenation and mono-methylation of nitroaromatic hydrocarbons with methanol as C1 and hydrogenation source, with water as the only by-product. A series of nitroaromatic hydrocarbons, including heterocyclic or sterically hindered derivatives, can be well tolerated and the catalyst could also be reused 4 times without losing significant reactivity. At the same time, the study of the Rh/MelCOF mechanism supports the hydrogen borrowing mechanism and puts forward the reaction pathway of azobenzene as an intermediate, which is better than the hydrogen transfer pathway from N-phenylhydroxylamine to aniline directly. This work not only expands the COF family but also provides an effective way to obtain mono N-methylated amines from nitroaromatic hydrocarbons, as well as the detailed mechanism of Rh/COF catalyzed tandem transfer hydrogenation and mono-methylation of amines.


Scheme 1.
Synthesis of N-methylamines by using methanol. Covalent organic frame materials are a kind of periodic and crystalline organic porous polymers connected by covalent bonds. Because of its good thermal and chemical stability, ordered pore structure, good crystallinity, and designability of the unit structure, it has become a research hotspot in recent years. [39][40][41][42][43][44][45] Among them, the Schiff base COFs are connected by C=N, and both N in the precursor and N in the Schiff base can be used to anchor metal atoms to form strong coordination bonds. Nowadays, COFs are used in Suzuki reaction [46,47], Sonogashira reaction [48], Diels Alder reaction [49], Michael addition reaction [50], and so on [46], but there are few reports on the application of COFs supported transition metal in N-methylation reaction. Specifically, when concerning the challenging tandem hydrogenation and N-mono-methylation of nitroaromatic hydrocarbons, especially when methanol served as both methylation and hydrogenation source, it undoubtedly requires that the transition metals should have the unique properties which can dehydrogenate methanol for reductive amination and further catalyze the Nmethylation of amines selectively, and the precious metals are normally one of the choices. As mentioned above, catalyst stability and recyclability are of great importance facing the current environmental and economical issues. So, in combination with the advantages of both COF materials and precious transition metal, developing high and efficient heterogeneous catalysts can not only facilitate the challenging tandem experimental operation but also can recover the catalyst, this definitely will open COF supported transition metal catalyst wide application in both academic and industrial. Therefore, we designed and synthesized two kinds of Rh-anchored COFs using melamine as the precursor, and reported for the first time the tandem reaction of transfer hydrogenation and N-mono-methylation of nitroaromatic hydrocarbons with methanol as the methylation source. In addition, the catalyst is suitable for nitroaromatic hydrocarbons with various reaction functional groups and can be recycled without losing reactivity.

Catalyst Preparation
Melamine (500mg) and piperazine-1,4-dicarbaldehyde (1139mg) were added to 60 ml DMSO at a molar ratio of 2:3 and ultrasonic until the precursor was completely dissolved. Then nitrogen protective condensation refluxed was carried out at 180 °C for 72 hours. Then the white sediment was collected and washed with ethanol, acetone, tetrahydrofuran, and dichloromethane respectively. Finally, vacuum drying at 90 °C for 8 h, and expressed as MelCOF-1. For comparison, MelCOF-2 was synthesized using melamine (500mg) and terephthalaldehyde (1074mg) at the same molar ratio of 2:3 and the same method. To prepare Rh-anchored COF, the RhCl3 methanol solution is introduced into the pre-synthesized COF with a metal content of 5wt% and stirred at room temperature for 24 hours. After the reaction, the excess unloaded RhCl3 was removed by Soxhlet extraction, and then dried in a vacuum at 80 °C for 12 h.

Catalyst Characterization
The crystal structure of the catalyst was recorded by X-ray powder diffraction using Cu-Kα radiation, with 2θ ranging from 5° to 80°. X-ray photoelectron spectroscopy (XPS) was carried out using AXIS Ultra DLD (Kratos, Britain), and the standard C1s peak is used as a reference for correcting the shifts. Field emission scanning electron microscope (FESEM) images and transmission electron microscope (TEM) images were performed to determine the morphology of the catalyst. These images were taken by JEOL JSM-6700 and JEM-2010 electron microscopes. After vacuum degassing at 90 °C for 15 hours, the adsorption-desorption isotherms of N2 samples were measured by Tristar3010 isothermal nitrogen absorption analyzer (Micromeritics, USA) at 77 K. The 1 H NMR spectrum is carried out on Bruker400 with reference to the use of tetramethylsilane by 400MHz in CDCl3. The 13 C SSNMR spectrum was recorded on the Bruker AVANCE III spectrometers.

Catalytic Tests
The methylation of nitroaromatic hydrocarbons is carried out in a pressure tube. The 3 mg catalyst is usually placed in the pressure tube together with 0.2 mmol nitroaromatics 2.4 mmol cesium carbonate and 2 ml methanol. And put it into the preheated oil bath pot, magnetic stirring heating for the corresponding time. After the reaction time is over, the pressure tube is removed and cooled to room temperature, and the solid catalyst is separated by a 0.22 μm filter membrane. The mixture was analyzed by GC or GC-MS and quantitatively analyzed by the normalization method.

Characterization and analysis of catalysts
Both COFs catalysts were synthesized by simple stirring, and the synthesis process was as shown in Scheme 2. Melamine was used as a monomer to react with piperazine-1,4-dicarbaldehyde or terephthalaldehyde in DMSO to obtain MelCOF-1 or MelCOF-2. Then the metal ions were introduced into the COFs by the strong coordination interaction between the metal ions and the N in the substrate of the COFs, and the Rh/MelCOF-1 or Rh/MelCOF-2 was obtained.

Rh
Scheme 2. Representation of the synthesis of Rh/MelCOF. The synthesized MelCOF and Rh/MelCOF were characterized by scanning electron microscope (SEM) and transmission electron microscope (TEM). Typical SEM images, as shown in Figure 1 a, b, d, e, revealed that MelCOF and Rh/MelCOF showed the irregular shape of 200nm, and the introduction of Rh did not destroy the structure of COF, but made the whole structure of COF more agglomerated, forming a compact state, resulting in the decrease of its morphology and size. In order to understand the crystal structure of the synthesized COF, powder X-ray diffraction (PXRD) was carried out. Figure 1c, 1f shows the PXRD diffraction spectra of the amorphous MelCOF and the corresponding Rh/MelCOF. As expected, since there is no regular nm-level morphology in SEM, and Rh riveted MelCOF is in ionic form, the diffraction peak of COF in the low angle range and the diffraction peaks of Rh are not observed. The relatively wide signal at 22° coincides with the reflection (001), which is attributed to the π-π accumulation between the ordered adjacent layers of COF sheets.
XRD pattern of as-synthesized MelCOF and Rh/MelCOF(c,f). The adsorption-desorption isotherms of MelCOF and corresponding Rh/MelCOF are shown in Figure 2. It was found that the Brunauer-Emmett-Teller (BET) surface areas of MelCOF-1 and Rh/MelCOF-1 were 357m 2 g -1 and 239m 2 g -1 respectively. The BET surface areas of MelCOF-2 and Rh/MelCOF-2 were 457 m 2 g -1 and 284 m 2 g -1 , respectively. In addition, according to the Barrett-Joyner-Halenda (BJH) method, the load Rh reduces the Vpore of MelCOF-1 and MelCOF-2 from 0.339 cm 3 and 0.827 cm 3 to 0.289 cm 3 and 0.592 cm 3 , respectively. The reduction of specific surface areas and pore volume can be attributed to the successful encapsulation of ionic rhodium in MelCOF channels after the introduction of Rh. The relatively narrow pore size distribution of Rh/MelCOF-1 and Rh/MelCOF-2 is concentrated between 1.5 nm-5.7 nm, and the pore size changes slightly compared with the carrier MelCOF, indicating that the existence of Rh does not block the cavity channel of Rh/MelCOF. In addition, the Rh/MelCOF-1 and Rh/MelCOF-2 supported by MelCOF materials are further analyzed by EDS. As shown in Figure 3, there exist S elements in addition to C and N of synthetic COF materials. Although a variety of solvents are used to wash the COF material carrier in the synthesis process, and Soxhlet extraction is used in the process of loading Rh, there is still a small amount of DMSO unwashed, so some S elements is detected. At the same time, the Rh element was also observed. It can be seen that Rh was uniformly loaded into MelCOF, but the content of Rh (1.32Wt%) in Rh/MelCOF-1 was significantly lower than that in Rh/MelCOF-2 (5.86Wt%).  Figure 4 (a) clearly shows the signals of Rh3d, N1s, C1s, and O1s. C1s is used as a reference to correct the binding energy of XPS analysis under 284.8 eV and used for analysis. The corresponding high-resolution C1s spectra of Rh/MelCOF-1 and Rh/MelCOF-2 in Figure 4 (b) show that the peaks centered by 287.0 eV and 287.5 eV follow the sp2-coordinated C=N bond on COF materials [51]. In addition, the two peaks fitted by the N1s spectra Figure 4 (c) of the two kinds of Rh/MelCOF also belong to sp2 hybrid nitrogen, corresponding to the environment of C-N=C and C-N, respectively. Then the most important thing is the Rh3d XPS spectrum. From Figure 4 (d), Rh/MelCOF-1 can see that the binding energy spectrum has two peaks in 309.6 eV and 314.1eV, which are Rh3d3/2 (Rh-Nx) and Rh3d5/2 (Rh-Nx) energy levels belonging to Rh (III) species. At the same time, the 305.9 eV satellite peak of Rh is also observed. There is also a corresponding peak in Rh/MelCOF-2. In addition, compared with Rh/MelCOF-1, the Rh/MelCOF-2 has more signal peaks of 317.0 eV and 312.1eV, which is attributed to the peak of rhodium chloride that is not strongly coordinated with N in the COF ligand. Thus it can be seen that the Rh species mainly exist in the form of Rh-NX in Rh/MelCOF-1, while in the Rh/MelCOF-2 catalyst, in addition to Rh-NX, there is rhodium chloride that filled in the MelCOF2 channel. Therefore, although the amount of rhodium chloride used in the synthesis process of the two COF catalysts is the same, the content of Rh on the actual load MelCOF-2 is higher than that of MelCOF-1, which also confirms the results of EDS characterization. The binding energy of rhodium chloride decreases from 317.0 eV and 312.1 eV energy to 314.6 eV and 309.7 eV respectively, which indicates the electron transfer from N to Rh and reveals the strong metal-carrier interaction between Rh and MelCOF. XPS spectra of the synthesized Rh/COFs.

Study on Catalytic performance
The catalytic applicability of two kinds of Rh/MelCOF-1 or Rh/MelCOF-2 was studied by investigating the single N-methylation of nitrobenzene as a model reaction ( Table 1). In the initial condition screening process, we used Rh/MelCOF-2 with a relatively high content of Rh load, and the substrates were 2 mmol nitrobenzene, 2 ml methanol and 1.2 eq base. With the increase of reaction time, the conversion of reactant 1a increased, while the selectivity of by-product 1e and 1f decreased, while the selectivity of main product 1c increased, so it was inferred that 1e and 1f were possible intermediates. The reaction temperature has a great influence on catalytic performance. When the reaction temperature increases from 90 to 130 degrees Celsius, the conversion rate of 1c is also greatly increased to 98%. At the same time, it is observed that only 1e and 1f are formed at 90 degrees Celsius, indicating that these two substances are preferentially formed under the catalysis of Rh/MelCOF, which provides an important idea for the kinetic analysis in the following article. In order to confirm whether the reaction was catalyzed by Rh or MelCOF in Rh/MelCOF, other comparision experiments were conducted and found that 1c was not formed in the reaction without catalyst, or only using RhCl3 or MelCOF2 as the catalyst. Then the effects of different bases on the reaction were evaluated at 130 °C for 16 h. Different from the effects of different kinds of bases in the existing literature, almost all inorganic bases had good selectivity, such as Cs2CO3, NaOH, KOH, KO t Bu, Na2CO3, K2CO3, and there was no methylation. The formation of 1d was mainly the selective competition between 1b and 1c. However, the conversion of organic alkali Et3N is low due to its weak basicity. For this reason, under the optimum reaction conditions of 130 °C, 16 h and Cs2CO3, the 1c selectivity is lower when using the Rh/MelCOF-1 than that of use Rh/MelCOF-2. Then, based on the optimized conditions of methylation and hydrogenation of nitroaromatic hydrocarbons using methanol as the source of methylation and hydrogenation, it is proved that Rh/MelCOF is generally applicable in the synthesis of N-monomethylamine substrates. As shown in Table 2, 2 mmol substrates, 3 mg Rh/MelCOF-2 catalyst and 2.4 mmol Cs2CO3 (1.2eq) are used for COF supported Rh catalyzed tandem hydrogenation and mono-methylation in 2 ml MeOH at 130 °C for 16 h. It can be seen that many nitro compounds with different functional groups have good tolerance, which proves their universal application in mono-N methylation. Among them, the substrates with para-electron donor groups have good reaction selectivity, and all of them are quantitatively converted into monomethylated derivatives (2c, 5c, 5c, 6c, 7c). However, the electron donor groups on the m and ortho-position will lead to the decrease of selectivity and even incomplete conversion, and the main competitive product is aniline aromatics, which may be attributed to the increase of spatial shielding of nitro functional groups. As a result, the transfer hydrogenation step is inhibited (3c). However, the conversion and selectivity of electron-withdrawing halogensubstituted derivatives at the para-position are poor, and the main competitive products are aniline aromatics, azobenzene, and azobenzene oxide homologs (8c, 9c, 10c). On the other hand, the reaction effect of the meta-electron-absorbing group is better than that of the para-substituent, the conversion can reach more than 99%, and the selectivity is also higher, which may be attributed to the synergistic effect of electron synergy and space effect (11c, 12c, 13c). On the other hand, the spatial effect of ortho-electron-absorbing halogen-substituted derivatives is stronger than that of electron synergism, which leads to a decrease in the yield of N-monomethylamine (14c, 15, 16c). However, heterocycles, biphenyl rings, and substrates with steric hindrance all have good reactivity (17c, 18c, 19c, 20c). Moreover, according to the reaction law, the effect of electronic synergy on the methylation and hydrogenation of nitroaromatic hydrocarbons using methanol as a methylation source is higher than the spatial effect, and heterocyclic compounds can counteract the effect of electronic synergy. These prove that the prepared Rh/MelCOF heterogeneous catalyst can obtain hydrogen directly from methanol resources and carry out the mono-N methylation, and highly selective synthesize the challenging N-mono-methylamine products via the tandem transfer hydrogenation and mono-metylation of amine. Then the reusability of Rh/MelCOF-2 catalyst was tested for N-monomethylation reaction of 1a (Figure 5a). The catalyst recycle ability test was carried out under the optimized reaction conditions, and when every single reaction was finished, the reaction samples were analyzed using GC. Then the residue was centrifugally precipitated and the reaction solution was extracted, afterward, methanol is added to the remaining solid-liquid mixture, and again the centrifugal precipitation is carried out which is repeated until no color showed according to TLC. In this case, the next catalyst recycle test could be carried out according to the standard procedure, the reaction mixture is loaded into the tube and carried out for the next run. As shown in Figure 5a, the catalyst can be successfully recycled 4 times without losing significant catalytic reactivity and product selectivity. In the fifth cycle, although the conversion still kept over 99%, the selectivity of 1c decreased to 49%, which is probably due to the decrease of catalyst performance and the decrease in the reaction rate from 1b to 1c.
Under the optimized reaction conditions, the reaction curve of p-methylnitrobenzene 2a showed that 2a was rapidly consumed within 4 hours to form 2c (Figure 5b). In the early stage of the reaction, the intermediate compound 2e gradually increased and reached the highest amount within 2 hours and then decreased gradually after 2 hours. Interestingly, another intermediate compound 2f continuously stays at a low level with the emergence of 2e. The intermediate amine 2b also slowly increased to the maximum in 4 h, which is consistent with the result that the starting material 2a was completely consumed at 4 h. Then the intermediates compound 2b, 2e was gradually transformed into 2c, which was consistent with the observed selectivity.  Then, to support the hydrogen borrowing mechanism and determine that the reaction intermediates are involved in the reaction, 1b, 1c, 1e, and 1f are used as the substrates for the reaction under the optimized conditions. It is found that 1b and 1e are all reaction intermediates, and once 1c is produced, it is difficult to reverse the reaction. According to the reported kinds of literature, it is speculated that the hydrogen transfer reduction of nitrobenzene to aniline occurs in two ways, and then N-methylaniline is synthesized by N-methylation of aniline. One way is that nitrobenzene is reduced to N-phenylhydroxylamine by intermediate nitrosamine with the help of hydrogen transfer, which is directly dehydrated to aniline. The other path is the reductive coupling of nitrosamine with N-phenylhydroxylamine to get azobenzene by oxidation of azobenzene, which is further reduced to aniline by 1,2-diphenylhydrazine. From the analysis of mechanical inquiry dynamics, the second path plays a major role in it. Then, as previously reported, aniline was dehydrogenated with methanol to form the intermediate N-phenylmethylimine, and then continued to be reduced to Nmethylaniline (Figure 6). Figure 6. Proposed mechanistic pathways for the transfer hydrogenation, reduction and N-monomethylation of nitrobenzene to aniline using methanol as a hydrogen source.

Conclusions
In conclusion, we successfully synthesized and fully characterized two kinds of novel Rh/MelCOF catalysts, and realized the highly selective synthesis of N-methylamine by only using methanol as the methylation and hydrogenation source, and nitroarene as the raw material. Two kinds of Rh/MelCOF carrier materials were prepared by a simple heating and stirring method, and the relatively narrow pore size distribution was concentrated in 1.5 nm-5.7 nm. Rh is anchored uniformly on the two COF materials. According to XPS, it is found that the strong coordination bond of MelCOF has a strong interaction with Rh, which makes the catalyst have the catalytic performance of carrier and rhodium chloride to carry out the mono-N-methylation of nitroaromatic hydrocarbons. A series of functionalized nitroaromatic hydrocarbons with electron-donating and electron-withdrawing substituents, including heterocyclic or sterically hindered derivatives, have been effectively converted to corresponding N-methylamines with good to excellent yields under low catalyst support using Cs2CO3. On the other hand, according to the study of the reaction path, the reaction path here is different from the previously reported mechanism, and the reaction path with azobenzene as an intermediate is faster and much more selective. With the help of Rh/MelCOF, it is widely used to realize the high-value conversion of nitroaromatic compounds to a variety of Nmethylamines. The synthesis process using methanol as methylation and hydrogenation source is more economical and green. Currently, we are focusing modify and improve the stability of these novel COF-supported heterogeneous catalysts and exploring their more applications in challenging organic transformations.