Magnetic composite of γ-Fe2O3 hollow sphere and palladium doped nitrogen-rich mesoporous carbon as a recoverable catalyst for C–C coupling reactions

In this article, palladated-magnetic nitrogen doped porous carbon was prepared from nano magnetic γ-Fe2O3 hollow sphere (h-Fe2O3) with high specific surface area and pore volume. To the purpose, initially h-Fe2O3 was prepared and covered with glucose via hydrothermal treatment with subsequent polymerization of organic shell. The polymerization of melamine-resorcinol–formaldehyde (MRF) was achieved in the presence of Cl-functionalized glucose coated h-Fe2O3 (h-Fe2O3@glu-MRF). Next, the prepared magnetic core–shell hollow sphere was palladated followed by carbonization to yield Pd@h-Fe2O3@C introducing more pores in its structure. The resulted compound, Pd@h-Fe2O3@C, was fully characterized, showing that carbonization process expressively increased the specific surface area. The resulted Pd@h-Fe2O3@C was successfully used for promoting C–C coupling reactions under mild reaction conditions as a heterogeneous catalyst and its activity was compared with some prepared control catalysts. This novel catalyst was magnetically separated simply by a magnet bar and recycled and reused at least in five consecutive runs, without considerable loss of its activity. It is note mentioning that, high recyclability with low Pd leaching are another gains of this protocol.


Result and discussion
Catalyst characterization. First, morphological characteristics of the Pd@h-Fe 2 O 3 @C including shape, size, and particle distribution envisioned by TEM, high resolution transmission electron microscopy (HRTEM) and selected area electron diffraction (SAED) pattern. TEM images displayed hollow-spherical morphology albeit with some amount of agglomeration and the diameter of the Pd nanoparticle was about 17 nm ( Fig. 2A,C). Moreover, it is clearly that the spheres are surrounded with gray coverage, confirming the successful incorporation of glucose and organic layer in the structure of the catalyst 22 . As can be seen, the HRTEM image displays the crystalline lattice structure inside the Pd NPs. The lattice fringe spacing was measured to be 0.20-0.28 nm (Fig. 2D,E), which can be attributed to the Pd (111) plane. Figure 2F represents its SAED pattern, indicating www.nature.com/scientificreports/ different planes, which matches with the XRD planes. From the histogram, the mean diameter was found that ∼16.1 nm with a standard deviation of 4 nm. The average diameter of the Pd nanoparticles should be equal to the distance (20 nm), estimated above. Furthermore, the average diameter size of Palladium NPs was calculated by employing Debyee-Scherrer equation, being about 19.5 nm. The variation in the mean diameter from one agglomerate to another is acceptable and may be attributed to the blurring of boundaries in the 3D structures. As visible, our synthesis method resulted in a selective deposition of the external Pd layer at the surface of the magnetic composite. Next, FTIR spectroscopy was applied to verify the formation of Pd@h-Fe 2 O 3 @C and the compounds prepared in the course of synthesis of Pd@h-Fe 2 O 3 @C. The FTIR spectra of the Pd@h-Fe 2 O 3 @C (a), h-Fe 2 O 3 @glu-MFR (b) and pure glucose (c) are depicted in Supplementary, Fig. S1. This spectrum (a) clearly showed the characteristic bands of h-Fe 2 O 3 , i.e. the strong absorption bands at 470-590 cm −1 , which can be assigned to Fe-O stretching and the strong band at 3463 cm −1 , which can be attributed to -OH groups 11 . The characteristic bands of glucose (c) can be listed as the bands at 3411, 2943 and 1461 cm −1 that can be assigned to the -OH functionality and -CH 2 stretching. The FTIR spectrum of h-Fe 2 O 3 @glu-MFR (b) exhibited the characteristic bands at 1604, 2850, 2942 and 3406 cm −1 can be due to the C=N, -CH 2 stretching and -OH functionality, which confirms the conjugation of organic layer. Moreover, the FTIR spectrum (b) showed the characteristic bands of magnetic core (h-Fe 2 O 3 ), implying that the magnetic core preserved its structure upon functionalization processes. The FTIR spectrum (a) of Pd@h-Fe 2 O 3 @C is significantly distinguished from others and the intensity of the characteristic bands became very weak. This observation of some of characteristic bands is quite expectable and can be due to the high temperature thermal treatment and carbon generation.
The structure and formation of γ-Fe 2 O 3 hollow sphere as well as palladium NPs was also studied by X-ray diffraction pattern of the catalyst (Fig. 2 25 . It is worth mentioning that the characteristic bands at 2θ = 81.7° was overlapped with that of Palladium NPs. Raman spectroscopy was also applied for the characterization of the catalyst (see Supplementary Fig. S2). The Raman spectrum of Pd@h-Fe 2 O 3 @C exhibited two bands at 1357 (D-band) and 1601 cm −1 (G-band), related to graphitic carbon, can confirm the graphitic nature of the catalyst. In greater detail, the D-band is indicative of the sp 3 configuration that can be attributed to presence of intrinsic defects and the G-band can be attributed to the graphitic carbon 26 . In this line, the I D /I G was calculated and measured to be 0.83, can confirm disordered graphitic structures or highly defective 27 .
To elucidate whether decoration of the surface of h-Fe 2 O 3 with carbon shell and palladium NPs can alter the magnetic property, Pd@h-Fe 2 O 3 @C was studied by room temperature vibrating sample magnetometer (VSM), and its magnetic features compared with that of h-Fe 2 O 3 (see Supplementary Fig. S3). As obvious, the maximum saturation magnetization (Ms) values of h-Fe 2 O 3 and Pd@h-Fe 2 O 3 @C were found to be 45.1 and 33.1 emu/g, respectively. Clearly, the M s value of the hollow Fe 2 O 3 nano sphere (45.1 emu/g) is higher than that of the catalyst www.nature.com/scientificreports/ which may be due to the incorporation of non-magnetic compounds and immobilization of palladium NPs on the surface of magnetic core. However, the hysteresis loops of the catalyst showed a paramagnetic behaviour without aggregation that can be easily separated from the reaction mixture using an external magnetic force. In Supplementary Fig. S4, the thermal stability of the (a) h-Fe 2 O 3 @glu-MFR, (b) h-Fe 2 O 3 and (c) Pd@h-Fe 2 O 3 @C were recorded using TG analysis and were compared. As shown, the h-Fe 2 O 3 possessed high thermal stability and exhibited only the weight loss below 200 °C that is representative of loss of water. The comparison of h-Fe 2 O 3 @glu-MFR with that of h-Fe 2 O 3 indicates that apart from the weight losses of magnetic core, an major weight loss at 450 °C can be detected that can be attributed to the degradation of organic layer. More detailed, the weight loss of about 19% can be due to the decomposition of MRF as an organic motif that indicates the successful formation of MRF in the structure of the catalyst. Considering the thermo gram of Pd@h-Fe 2 O 3 @C, it can be concluded that this sample exhibited significantly higher thermal stability compared to that of others, confirming the successful carbonization.
In the following, the effect of the carbonization of organic shell on the textural properties of the catalyst was studied. To this purpose, the N 2 adsorption-desorption isotherms of Pd@h-Fe 2 O 3 @C and h-Fe 2 O 3 were recorded and depicted in Supplementary Fig. S5. As shown, the isotherms of two samples are distinguished. The shape of h-Fe 2 O 3 exhibited type II isotherm, while Pd@h-Fe 2 O 3 @C showed type IV with H3 hysteresis loops 24 . To further verify this issue, the specific surface area of two samples were calculated and compared. The specific surface area of the catalyst was calculated to be 426 m 2 g −1 which was higher than that of h-Fe 2 O 3 (53 m 2 g −1 ). This can indicate the porous nature of Pd@h-Fe 2 O 3 @C. The total pore volume of two samples were also compared (see Supplementary Fig. S5). More precisely, this value for Pd@h-Fe 2 O 3 @C was much higher than that of h-Fe 2 O 3 , indicating the carbonization of organic layer resulted in the formation of pores. Moreover, the pore size distribution curves of the h-Fe 2 O 3 and Pd@h-Fe 2 O 3 @C were obtained by the BJH method using the pore volumes in the measurement of N 2 desorption isotherms. As is evident from pore size distribution result of Pd@h-Fe 2 O 3 @C, two types of pores with mesoporous (2 nm) and micropores (8.9 and 11 nm) were clear. Nevertheless, compared to the Pd@h-Fe 2 O 3 @C, the microporous size uniformity has increased and appeared at 8.9 and 11 nm, presumably due to the carbonization of organic layer.
Finally, ICP-AES was exploited for measuring the content of Fe and palladium NPs in the catalyst. To prepare the sample for the analysis, a known quantity of Pd@h-Fe 2 O 3 @C was digested in a mixture of concentrated HCl and HNO 3 solution. The obtained extract was analysed and the results confirm that the loading of Fe and palladium were 2.47 and 0.075 mmol g −1 , respectively.

Investigation of the catalytic activity.
To assess the catalytic activity of this heterogeneous system, the Pd@h-Fe 2 O 3 @C was utilized as a recyclable catalyst in Suzuki coupling reaction. Initially, the reaction of iodobenzene 1 and phenylboric acid 2 was selected as a model substrate. The reaction was conducted in the present of Na 2 CO 3 and Pd@h-Fe 2 O 3 @C (0.5 mol%) at 75 ℃ in EtOH condition. After 1 h, the nature of the reaction mixture quickly changed to a dark viscous solid. After purification by column chromatography, the biphenyl 3a was afford in 35% yield. Ultimately, the experimental results established that the biphenyl could achieved with 95% yield when the model reaction was carried out in the present of Pd@h-Fe 2 O 3 @C (0.5 mol%) with Na 2 CO 3 at 75 ℃ in water/EtOH. Further, to determine its scope by applying various aryl halides, a range of reactions was carried out under the optimal reaction conditions. As shown in Table 1, a series of aryl halides with electron-donating and electronwithdrawing group and phenyl boronic acid were used to the reaction under optimized conditions. Eventually, the Pd@h-Fe 2 O 3 @C efficiently catalyzed the coupling reaction between aryl halides with phenylboronic acid and biphenyls were attained in high to excellent yields after purification. In detail, the electronic effect of the substituents was generally found have no considerable influence since aryl iodides bearing donor-and acceptor substituents reacted with phenylboronic acid to afford the expected coupled products in excellent yields. Worthy to mention that, when iodobenzene was substituted by bromobenzene, the reactions needed longer reaction times for being completed. Notably, the scope of this methodology was found not to be operative to chlorobenzene.
Furthermore, the performance of Pd@h-Fe 2 O 3 @C catalyst was tested for the Sonogashira coupling reaction using bromobenzene 1 and phenylacetylene 4 as a model substrate. As shown in Supplementary Table S2, the optimization reaction conditions were investigated in different solvents, temperatures, bases, and in the presence of various amount of the Pd@h-Fe 2 O 3 @C catalyst. It was found that, when the model substrate was applied in the presence of Na 2 CO 3 (2.0 mmol) at 50 ℃ using Pd@h-Fe 2 O 3 @C (0.35 mol%) as catalyst, the yield of product could reach 98% after separation. Using the optimized reaction conditions, the scope and generality of this method were exemplified in the reaction of aryl halides 1 and phenyl/aliphatic acetylenes 4 using Pd@h-Fe 2 O 3 @C catalyst, and the outcomes are presented in Table 2. The cross-couplings of phenyl/aliphatic acetylenes with aryl iodides bearing electron donating groups, -OMe, -Me and -COMe gave products in satisfactory yields. Moreover, arylbromides and chlorides are efficiently reacted as substrates in this process, though, the reaction of aryl chlorides needed longer reaction times for being completed. Noticeably, propargylalcohol as an aliphatic acetylene was also fruitfully coupled to arylhalides with satisfactory yields. Nevertheless, the highest yields were obtained for aryl acetylene. Noteworthy, all compounds are known and some were identified by comparing physical properties through FTIR and melting point analyses.
Next, the chemical state-catalytic activity relationship was studied. Initially, the result of loading of Pd on the support was studied. In this regard, apart from the catalyst two more samples with different loading of Pd, i.e. Pd@h-Fe 2 O 3 @C (Pd 2 and 3 wt%). The outcomes showed that the loading of Pd meaningfully influence the catalytic activity and use of lower content of Pd is more effective. Then, the influence of each component in the structure of the catalyst to the catalysis was studied. For this purpose, several control catalysts, including,  Table S3, entry 1, Pd@h-Fe 2 O 3 is not an active catalyst and the product was obtained only 35%. Then, it was studied whether introducing of Glu shell is able to improve the catalytic activity. In this line, Pd@h-Fe 2 O 3 @glu was provided and its catalytic activity was examined and the product was obtained with 50% yield (Table S3, entry 4). This results approved that Glu can somewhat improve the catalytic activity. For further revealing the key role of Glu in the catalysis, two control catalysts, Pd@h-Fe 2 O 3 @MRF and Pd@h-Fe 2 O 3 @MRF-C was prepared, in which the Glu was not present in the structure of the catalyst and the resorcinol-formaldehyde-melamine polymer was adjusted on the surface of   (Table S3, entries 2 and 3). Based on the comparison of the results, the contribution of Glu was confirmed. In details, donor-and acceptor substituted it was found that Glu component not only influenced the catalytic activity, but also improved the separation and reusability of the catalyst. More exactly, the separation of the catalyst was simpler and more efficient than that of Pd@h-Fe 2 O 3 @MRF and Pd@h-Fe 2 O 3 @MRF-C, while, the ICP examination of these samples approved the higher loading of Pd NPs in the catalyst. Approving the role of melamine as a nitrogen source, the effect of N-precursor by creating another control samples, Pd@h-Fe 2 O 3 @glu-RF, Pd@h-Fe 2 O 3 @glu-RF-C in which melamine was omitted in the structure of polymer was clarified and compared with Pd@h-Fe 2 O 3 @glu-MRF and catalyst. As tabulated, melamine as an N-rich precursor has the ability to produce product with the highest catalytic activity (Table S3, entries 5-8). To elucidate the effect of polymer's type on the structure of the catalyst and Pd loading, the specific surface area of these samples were compared and showed that this value decreased in the following order: Pd@h-Fe 2 O 3 @C (426 m 2 g −1 ) > Pd@h-Fe 2 O 3 @glu-RF-C (156 m 2 g −1 ) > Pd@h-Fe 2 O 3 @glu-MRF (42 m 2 g −1 ) > Pd@h-Fe 2 O 3 @glu-RF (33 m 2 g −1 ). By these results the effect of the N-rich carbon precursor on the content and specific area of the carbon coated h-Fe 2 O 3 was confirmed. Further, the effects of composite components on the loading and leaching of Pd NPs in catalyst as well as control samples were measured via ICP analysis and studied, Furthermore, the efficiency of this protocol and the catalytic performance of Pd@h-Fe 2 O 3 @C for catalyzing the Suzuki and Sonogashira model reactions were compared with those of some previously reported catalytic methodologies to disclose the merits of this procedure (Table S4). As obvious, Pd@h-Fe 2 O 3 @C resulted in desired product in higher or comparative yields. However, compared to all cases, Pd@h-Fe 2 O 3 @C led to the product in shorter reaction time. Moreover, this protocol does not required any harsh reaction condition, inert atmosphere or toxic solvent 28-33 . Catalyst recyclability. To explain whether Pd@h-Fe 2 O 3 @C can be considered as a reusable catalyst, the recycling of that catalyst was carried out for the Sonogashira model reaction and the results exhibited in Supplementary as Fig. S6. As shown, the catalyst was subjected to five successive runs. Noteworthy, up to the fourth reaction run, only a reasonable decrease was detected, subsequently a more obvious loss of activity after the fifth run and the yield of product reached to 44%. The recycling could have also been hampered by partial structural damage to support. There was also observed increase in mass of the solid due to the presence of the catalyst, solid base, and salt making reproducibility difficult after every run. In addition, washing with water after separation reduced the activity of the catalysts, because the pore size of the surface of mesoporous carbon shell is smaller than palladium nanoparticles. Filtering also leads to loss of some Pd NPs as an active sites. However, a hot filtration test proved the heterogeneous nature of the supported catalyst and without a measurable homogenous contribution. Table S3 gives the results of catalyst leaching in the Sonogashira coupling reaction of aryl halide and phenyl acetylene. According to the ICP-AES analysis, the Pd content of the heterogeneous catalyst was determined to be 0.075 mmol/g −1 . The percentage of Pd leaching/Pd loading was around 4.6%. The soluble leached Pd could likely be responsible for catalysis in this reaction. Pd leaching correlates significantly with the progress of the reaction, the nature of the starting materials and products, solvent, base, and atmosphere. These results agree with Shmidt and Mametova's observation of the oxidative attack of the halide to the metal crystallites, yielding directly Pd(II) in solution 34 . Generally, the metal ions are probably first reduced in the presence of base, which causes some leaching, which is followed by oxidative addition of aryl halide and substantial leaching 35 . To consider the effects of reusing on the morphology of the catalyst, FTIR spectra and TEM analysis of the recycled catalyst were recorded and compared with that of the fresh catalyst. The FTIR spectrum of the recycled Pd@h-Fe 2 O 3 @C demonstrated the specific bands of the fresh Pd@h-Fe 2 O 3 @C (Fig. 3a). Nevertheless, some difference between two spectra were observed, that can be due to the disposition of organic substances on the surface of the catalyst. The TEM image of the recycled Pd@h-Fe 2 O 3 @C demonstrated some of the agglomerated nanoparticles, which can be due to the magnetic nature of nanoparticles. However, the spherical structure of the nanoparticles is preserved (Fig. 3b). Furthermore, the ICP-AES analysis of the filtrate exhibited Pd and Fe content 0.0031 and 0.19 mmol g −1 , respectively.

Synthesis of the catalyst. Synthesis of nanomagnetic Fe 2 O 3 hollow sphere (a).
The palladated-encapsulated h-Fe 2 O 3 was fabricated through in situ polymerization. The procedure included the preparation of magnetic core, synthesis of prepolymer solution, and the formation of palladated-magnetic carbon shell. Nanomagnetic Fe 2 O 3 hollow sphere as a magnetic core, was prepared through solvothermal method 11

Synthesis of h-Fe 2 O 3 @glu core-shell (b).
The resulting h-Fe 2 O 3 @glu core-shell was prepared by previous method with little modification, accordingly 36 . Initially, h-Fe 2 O 3 (1 g) well dispersed in deionized water (40 mL) by using ultrasonic irradiation (power 100 W) for 15 min. Afterward, glucose (6 g) was introduced in the prepared magnetic suspension and the resulting mixture was transferred into a Teflon-lined stainless steel autoclave (150 mL). Then, the container was closed and maintained at 200 °C for one day. At the end of the hydrothermal treatment, the reactor was cooled down to room temperature and the product 2 was collected by a magnet bar, rinsed with EtOH, centrifuged for five times and dried in oven at 80 °C.

Synthesis of h-Fe 2 O 3 @glu-MFR (d).
The pre-polymer solution was prepared by mixing formaldehyde (10 mL) and melamine (2 g) in distilled water (10 mL), according to previous method 38 . Next, the pH of the mixture was adjusted to 8.5-9.0 by adding TEA and stirred vigorously at 70 •C. When it became clearly transparent, another melamine (1 g) was added and stirred till it was dissolved completely. Similarly, water (10 mL) was added and stirred till the solution was transparent absolutely. On the other, h-Fe 2 O 3 @glu-Cl (0.5 g) well dispersed in water (20 mL) and stirred for half time. The magnetic suspension was adjusted to pH 4.5-5.0 with 15.0 wt.% acetic acid solution. The prepared solution was added into the magnetic suspension under stirring condition, dropwisely. After that, resorcinol was added to the prepared cross-linking with the M-F pre-polymer. After all of the pre-polymer was added, ammonium chloride (5 wt.%) as a nucleating agent was added into the solution, then it was stirred at 60 •C for 90 min. The pH of the mixture was adjusted to 9.0 with TEA solution, which completed the reaction. Then the resultant microcapsule 4 was magnetically collected, washed with EtOH until pH 7 was reached. The wet powders were dried in a vacuum oven at 100 •C overnight. To this propose, Pd@h-Fe 2 O 3 @glu-MFR (5 g) was placed in a quartz container and heated up to 450 °C for 1 h under argon flow and heating rate of 30 °C min −1 . The sample was held at this temperature for 6 h. After cooling to room temperature, the product 5 (3 g) was washed and recovered as a black powder. The schematic processes of synthesis of the catalyst are depicted in Fig. 4.  . After finishing the reaction, Pd@h-Fe 2 O 3 @C was magnetically isolated, washed with EtOH repeatedly and dried in an oven at 60 °C for 6 h. Finally, the filtrate was extracted with diethyl ether for three times and then, the organic layer was washed with deionized water and dried over anhydrous Na 2 SO 4 and purified using column chromatography over silica gel.

Immobilization of palladium NPs on h-Fe
Suzuki reaction. A mixture of aryl halide (1.0 mmol), aryl boronic acid (1.2 mmol), Na 2 CO 3 (2.0 mmol) and Pd@h-Fe 2 O 3 @C (0.5 mol%) was prepared in the mixture of water/EtOH (1:1, 5 mL) and was heated at 75 °C in an oil bath, subsequently. The solvent for monitoring reaction process by TLC was n-hexane/ethyl acetate, 7:3. Then, the mixture was cooled to ambient temperature and the catalyst was separated by a magnetic field, washed with ethanol for three times and dried in oven at 60 °C for 6 h. After that, he solvent was removed and the product was extracted with 10 mL of n-hexane. To achieve corresponding biaryls, the product was purified by column chromatography over silica gel.
Characterizations of some products.