Graphene-supported Pd catalyst for highly selective hydrogenation of resorcinol to 1, 3-cyclohexanedione through giant π-conjugate interactions

The selective hydrogenation of resorcinol (RES) to 1, 3-cyclohexanedione (1,3-CHD) without the addition of alkali is a big challenge. In this article, a novel reduced graphene oxide (rGO) supported Pd catalyst was prepared through co-reduction method, over which we obtained 99.9% of resorcinol conversion and 94.2% of the ever-reported highest 1,3-cyclohexanedione selectivity at 25 °C in only CH2Cl2 solvent. The excellent selectivity was contributed to the strong π-π and p-π interactions between the graphene nanosheet and the benzene ring as well as hydroxyl in RES molecule. The followed adsorption experiment and Raman analysis also showed the existence of aromatic graphite structures in rGO, which exhibited stronger adsorption towards RES than towards 1,3-CHD.

Scientific RepoRts | 5:15664 | DOi: 10.1038/srep15664 benzene ring in phenol by hydrogenation 13 , the benzene ring in RES is more reactive as it is activated by one more adjacent hydroxyl groups through electron-donating effect, which would make the energy barrier of the addition of the first H 2 molecule be much closer to that of the addition of the remaining two H 2 molecules. However, it is much more difficult to obtain 1,3-CHD, the one-molecule H 2 addition product from RES. For example, Wang et al. 14 reported that C 3 N 4 -supported Pd catalyst showed better adsorption of phenols, over which 99% of selectivity toward cyclohexanone were obtained during phenol hydrogenation. While using the same catalyst for RES hydrogenation, however, 1,3-CHD was hard to be produced. Actually, to achieve a high yield of 1,3-CHD in an industrial process, the addition of an alkali or organic amine is always essential and the reaction may follow another pathway 4 .
Although the exact hydrogenation mechanism remains incompletely understood, modern research indicates that the surface properties and the spatial structure of the support perform significant functions in enhancing the selectivity of a reaction 15,16 . During hydrogenation of the aromatic ring over a heterogeneous catalyst, it is believed that H 2 molecules are decomposed into atoms over the metal active cites, while aromatic substrates are adsorbed onto the supports through hydrogen bonds, electrostatic effects and other interaction forces [15][16][17] . By this means, the aromatic ring is exposed to and attacked by the adsorbed hydrogen atoms and the hydrogenation reaction occurs.
The applications of graphene in materials science, condensed-matter physics 18 , and catalysis science 19 have been rapidly developed since its first discovery by Geim et al. in 2004 20 . Graphene is considered as a new-generation carbon support material because of its excellent physical and chemical properties 21 . Pristine graphene nanosheets generally present large surface areas, hydrophobicity, easy chemical functionalization, and stable structure 22,23 . Numerous theoretical studies have identified the presence of π -π interactions between graphene and aromatic rings [24][25][26][27] . For example, Wang's NMR and adsorption experiments 28 showed that phenol adsorption increases sharply after the reduction of graphite oxide. Molecular simulations performed by Björk's group 29 also confirmed the existence of π -π stacking interactions between graphene nanosheet and aromatic compounds. Inspired by these findings, herein we report the application of graphene supported Pd as a catalyst for the hydrogenation of RES. Since there is strong π -π conjunctive interaction force between graphene nanosheet and RES, and such interaction force will disappear when RES loses its aromatic character after the addition of the first H 2 , the intermediate product with only one molecule H 2 being added, i.e. the enol isomer of 1,3-CHD, is desorbed from graphene surface and in the end a high selectivity to 1,3-CHD is achieved. The graphene supported Pd catalyst can be reused several times after simple washing with conventional solvents.

Results and Discussion
We used the Hummers method [30][31][32][33] to produce graphene oxide and then reduced it together with PdCl 2 by the addition of NaBH 4 to obtain stable Pd/graphene (rGO) catalyst containing 3 wt% Pd loading. Catalytic hydrogenation of RES was conducted in a nonaqueous solution, i.e. CH 2 Cl 2 , as shown in Table 1. It can be seen that Pd/rGO catalyst exhibited much higher selectivity to 1, 3-CHD than the other Pd catalysts used in this work: in the experimental ranges, most of the selectivity to 1,3-CHD was more than 80% with a high conversion of RES over Pd/rGO catalyst. Typically, when the reaction took place at 25 °C and 1 MPa H 2 , 1,3-CHD selectivity was as high as 94.2 % with 99.9% conversion of RES after 3 h in the presence of Pd/rGO (entry 3), which is the highest selectivity that has been ever reported. In  comparison of Pd/AC catalysis, the selectivity to 1,3-CHD was no more than 30% (entry 8) at the same reaction conditions. Pd/SiO 2 and Pd/MWCNT catalysts also showed very low selectivity to 1,3-CHD (entries 9 and 10). Since the carbonyl group in 1,3-CHD is not stable under hydrogen environment, high temperature, high pressure or long reaction time could lead to decrease of the selectivity. For instance, 1,3-CHD selectivity was as high as 93.2% over Pd/rGO at 25 °C (entry 2), but dropped to 83.5% at 40 °C (entry 5). 1,3-CHD selectivity decreased from 86.2% (entry 5) to 54.5% (entry 6) at 40 °C as the hydrogenation pressure varied from 0.1 MPa to 2 MPa. With the reaction time varied from 1 h to 3 h, the selectivity to 1,3-CHD also decreased at different temperatures and pressures. The Pd/rGO catalyst can be reused for at least five times with stable catalytic performance after filtration and washing with CH 2 Cl 2 , at which the selectivity to 1,3-CHD still remained 93.8% with 99.9% conversion of RES (entry 7). Comparison of the XRD (Fig. 1) patterns and TEM (Fig. 2) images of the Pd/rGO catalyst before and after five reaction cycles shows no marked difference in morphology, Pd particle size or distribution. Although some reports argued that unsaturated bonds in graphene can probably be reduced in a hydrogen atmosphere 34 , our Raman spectra ( Figure s1) reveal that the graphene structure in the catalysts is not damaged because the reaction conditions used in this work are relatively mild.
GC-MS ( Figure s2) detected the main by-products as 3-hydroxy-cyclohexanone, 1,3-cyclohexanediol, and cyclohexanone. Cyclohexanone is mainly produced from the dehydration of 3-hydroxy-cyclohexanone and further addition of 1 mol of hydrogen. The speculated reaction path of RES hydrogenation is shown in Fig. 3.
It is very interesting to note that in Wang's research 14 , hydrogenation of RES over Pd/C 3 N 4 produced 99% of cyclohexanone other than 1,3-CHD, while Makowski et al. 9 reported 70% yield of m-hydroxy-cyclohexanone and 30% yield of cyclohexanediol by using another hydrophilic carbon-loading Pd as a catalyst. Two obvious differences may be observed between their work and ours. The first difference involves the solvent used. They preferred to utilize water as a solvent whereas we selected only CH 2 Cl 2 . If other solvents were applied in our reaction system, such as the polar protic water or methanol, or even the polar aprotic acetonitrile or DMSO, the reaction may either not occur or occur with very low activity and selectivity (Table s1). The phenomenon is quite identical to those of Han's work, in which a success catalytic process was developed for the selective hydrogenation of phenol to cyclohexanone in only nonpolar solvents like CH 2 Cl 2 and supercritical CO 2 12 . The second difference involves the polarity of the support. In Wang's 28 and Makowski's work 9 , numerous N atoms within the C 3 N 4 support and O atoms within hydrophilic carbon may be observed, and these atoms may form hydrogen bonds with hydroxyl and carbonyl groups and interfere with the hydrogenation reaction. In our work, rGO originating from the reduction of graphene oxide by NaBH 4 was used as a support. The molecular formula of this support is C 6 H 0.75 O 0.14 (entry 1, in Table s2) with the oxygen content of around 3 w%, indicating that most of the polar functional groups have been removed such that the resultant graphene nanosheets are completely hydrophobic. Different hydrogenation products were obtained due to the discrepancies of the solvent and support, indicating that different catalytic mechanism was followed [35][36][37][38][39][40] . Wang 14 attributed the excellent selectivity of the Pd/C 3 N 4 catalyst to phenol forming strong interaction force with nitrogen in the support through hydrogen-bond interactions while cyclohexanone cannot. As a consequence, the intermediate product cyclohexanone is more likely to desorb out from the support and thus the excessive hydrogenation is prevented. In our case, the hydrophobic support of rGO can hardly form hydrogen bonds. We therefore suppose that the excellent selectivity observed is due to the π -π and p-π conjugation interactions of giant π bond on the graphene nanosheets with the aromatic ring and hydroxyl in RES. According to this speculation, we can easily determine that the polar solvents we have tested are not appropriate for the selective hydrogenation of RES because of their negative impacts on π -π and p-π interactions. To further verify this thought, we prepared rGO-E and rGO-H by reducing graphene oxide using ethylene glycol 41,42 and hydrazine hydrate 43 , respectively, as reducing agents. Both the supports supported Pd catalysts exhibited poor performance for RES hydrogenation because rGO-E (with formula of C 6 H 2.45 O 1.39 , in Table s2) contains more than 23 w% of hydrophilic oxygen atoms as ethylene glycol cannot reduce graphene oxide completely and hydrazine hydrate introduces 4 w% of hydrophilic nitrogen atoms into the support (Table 2s), which will also interfere in the π -π and p-π interactions between graphene surface and the substrate.
Further proof comes from the adsorption tests of RES and 1,3-CHD on different supports. As shown in Figure s3 and s4, both the adsorption of RES and 1,3-CHD on rGO reached equilibrium after 20 min, and the equilibrium adsorption capacities toward RES and 1,3-CHD were 0.072 and 0.039 g·g −1 , respectively. The corresponding selectivity coefficient was 1.89-2.02 (Table 2), which means that RES adsorption occurs more extensively than 1,3-CHD adsorption. Although SiO 2 showed large surface area, and its adsorption capacities to RES and 1,3-CHD were larger than those of rGO, the corresponding

Table 2. Surface character of various catalysts and their adsorption properties towards resorcinol and 1,3-CHD. Note:
The saturated adsorption amount is estimated as the maximum adsorption amount of the solute with the increase in its concentration in CH 2 Cl 2 solvent. a Metal dispersion was estimated through CO-TPD by assuming the adsorption stoichiometry of Pd/CO = 1.5 47 .
Scientific RepoRts | 5:15664 | DOi: 10.1038/srep15664 selectivity coefficient approached 1.00, indicating no adsorption selectivity between RES and 1,3-CHD was observed. AC also showed large adsorption capacities to RES and 1,3-CHD, but the selectivity coefficient was only 1.16-1.21. In contrast, MWCNT showed favorable adsorption selectivity toward RES (selectivity coefficient 1.58-1.65, in Table 2) which is comparable to that of rGO, indicating that the giant π bonds of graphite structure in these two supports have strong interaction forces with RES. It is well known that a heterogeneous catalytic reaction starts with the adsorption of the substrate onto the support and followed by reaction at the catalytic active center. Our adsorption experiment shows that graphite structures exhibit stronger adsorption toward RES due to the π-π and p-π conjugate interactions between them, which means that RES is superior to 1,3-CHD in occupying active sites. After addition of 1 mol of hydrogen to RES, 1,3-CHD (enol isomer) was generated, and its less interaction force with graphene surface allows it to leave away easily. Thus, excessive hydrogenation is prevented and the selectivity toward 1,3-CHD is enhanced. As a result, Pd/rGO showed the highest selectivity to 1,3-CHD (94.2%, entry 3) and Pd/SiO 2 showed the lowest selectivity to 1,3-CHD (21.1%, entry 9) since SiO 2 shows no difference in the adsorption of RES and 1,3-CHD. The catalytic mechanism is illustrated as in Fig. 4.
The graphitization degree of a carbon material is generally characterized by the I G /I D value in the Raman spectra in which I G is the intensity of G band around 1575 cm −1 corresponding to sp 2 -bonded carbon atoms in a hexagonal graphitic ring, and I D is the intensity of D band around 1350 cm −1 corresponding to the vibrations of sp 3 -bonded carbon atoms of defects and disorder. As shown in Fig. 5, the I G /I D order of the carbon materials is: MWCNT > rGO > AC, indicating that MWCNT would possess the highest adsorption selectivity coefficient and be the best support for the selectivity hydrogenation of RES to 1,3-CHD, which is somewhat contradict to our experimental data. As the particle sizes and dispersion of Pd in rGO and MWCNT do not vary so much that may affect the selectivity (Table 2), we deduce the main reason may be that MWCNT containing the bending graphite surface weakens the interaction of its giant π bonds with other aromatic compounds, resulting in a lower adsorption selectivity to RES and lower reaction selectivity to 1,3-CHD than those of rGO 44,45 . Other carbon materials with more perfect graphite structure 26,46 are also under developing for the selective hydrogenation of aromatic compounds in our lab.

Conclusions
In conclusion, Pd/rGO catalyst was prepared by co-reduction of graphene oxide and PdCl 2 with NaBH 4 . The as-prepared novel catalyst showed high activity and selectivity for the hydrogenation of RES to 1,3-CHD, which could be attributed to the π -π or p-π strong interaction between the giant π bonds on the graphene nanosheets and the aromatic ring and hydroxyls in RES.

Methods
Chemicals and Materials. All reagents were obtained from Sino-pharm Chemical Reagent Co. Ltd, except that PdCl 2 was purchased from Shanghai Jiuling Chemical Co. Ltd.

Synthesis of graphene oxide (GO).
Graphene oxide was synthesized from natural graphite powder by using a modified Hummers method consisting of two steps of oxidation. In the first pre-oxidation step, concentrated H 2 SO 4 (40 mL), K 2 S 2 O 7 (8.4 g), and P 2 O 5 (8.4 g) were added into a 500 mL round-bottom flask and maintained at 80 °C for 4.5 h. Then the mixture was cooled to room temperature, diluted with deionized water, left overnight, vacuum-filtered, and washed with deionized water (1.6 L) to obtain the pre oxidized material. In the second oxidation step, concentrated H 2 SO 4 (230 mL) and the pre oxidized material were added into a 1 L three-necked round-bottom flask and chilled to 0 °C. KMnO 4 (60 g) was added carefully under continuous stirring to keep the reaction temperature below 10 °C for 30 min. Afterwards, the reaction temperature was gradually increased to less than 35 °C and maintained for 2 h. The mixture was diluted with deionized water (0.5 L) and stirred for 2 h, successively diluted with an additional deionized water (1.5 L), drop wisely added with H 2 O 2 (25 mL), and left undisturbed for 4 days. The precipitate was washed with HCl (1 mol·L −1 ) and centrifuged for at least three recycles to remove residual metal oxides, and repeatedly washed with deionized water and centrifuged until the filtrate became neutral. Finally, the brown mixture dispersion in water was sonicated for 30 min, centrifuged, and freeze-dried to obtain the final GO.
Synthesis of Pd/rGO. 1 g of GO was dispersed through sonication in 200 ml of deionized water. 52 mg of PdCl 2 was added and the mixture was rigorously stirred for 1 h. After that, 100 mL of NaBH 4 aqueous solution (7.5 g·L −1 ) was drop wisely added in and the mixture was incubated at 80 °C for 4 h. The resulting Pd/rGO catalyst was then filtrated and vacuum dried for use. The Pd content in the catalyst was determined to be 3.1 wt% by Atomic Absorption Spectrometer (AAS, Agilent AA240).
Synthesis of Pd/rGO-E. 1 g of GO was dispersed through sonication in 200 ml of ethylene glycol-water solution (3:2, v/v). 52 mg of PdCl 2 was added and the mixture was rigorously stirred for 1 h. After that, the mixture was incubated at 120 °C for 4 h. The resulting Pd/rGO-E catalyst was then filtrated and vacuum dried for use. The Pd content in the catalyst was determined to be 3.0 wt% by AAS.
Synthesis of Pd/rGO-H. 1 g of GO was dispersed through sonication in 200 mL of 1 w% N 2 H 4 aqueous solution. 52 mg of PdCl 2 was added and the mixture was rigorously stirred for 1 h. After that, the mixture was incubated at 95 °C for 8 h. The resulting Pd/rGO-H catalyst was then filtrated and vacuum dried for use. The Pd content in the catalyst was determined to be 3.1 wt% by AAS.

Synthesis of Pd/X (X = activated carbon (AC), multi-walled carbon nanotube (MWCNT), and
SiO 2 ). Pd/X was prepared by wetness impregnation method. Taking Pd/AC as an example, 1 g of AC was wetness impregnated by PdCl 2 acidic aqueous solution containing 5 mg·mL −1 of Pd (II). The impregnation process was repeated for several times till the calculated Pd content in the final catalyst reached ca 3 wt%. The solid was finally vacuum dried at 120 °C and reduced by hydrogen at 350 °C for 4 h to obtain the final Pd/AC catalyst. The Pd contents in Pd/AC, Pd/MWCNT and Pd/SiO 2 catalysts were 3.1 wt%, 3.0 wt% and 2.9 wt%, respectively, determined by AAS.
Characterization. The X-ray powder diffraction (XRD) patterns of the catalysts were measured with a Thermo X'TRA X-ray diffractometer using CuKα radiation (40 kV, 40 mA). Transmission electron microscopy (TEM) images were obtained using a Tecnai G2 F30 S-Twin instrument (FEI Co.) operating at an accelerating voltage of 300 kV. Raman spectra were collected on a Jobin-Yvon LabRam HR 800 Raman spectroscope equipped with a 514.5 nm laser source. BET specific surface areas and pore structures were measured by pulsed nitrogen adsorption-desorption method at − 196 °C using an ASAP 2010 instrument (Micromeritics Instrument Co.). Pd content in a catalyst was determined by Atom Absorption Spectroscopy AA240 (Agilent Technologies). The elemental composition of rGO was determined by Leco-Chns932 (LECO Co.). The dispersion of Pd was obtained by CO chemisorption method, which was carried out at 40 °C on a Quantachrome Autosorb-1/C chemisorb apparatus. Prior to measurements, the pre-reduced catalysts were reduced in situ for 2 h at 450 °C in H 2 . The metal dispersions were estimated based on an adsorption stoichiometry Pd/CO = 1.5 according to Han et al.'s work 47 , which were close to those obtained from TEM 36 . Typical procedure for hydrogenation of resorcinol. The reaction was carried out in a stainless batch reactor (20 mL) with a magnetic stirrer. In a typical procedure, resorcinol (0.027 mmol), catalyst (50 mg, ca 3 wt% of Pd), and 5 ml of CH 2 Cl 2 was added in the reactor. After purging 3 times with H 2 , the reactor was heated and pressurized with H 2 to the desire temperature and pressure to initiate the reaction. After reaction the reactor was placed in ice water to quench the reaction and the products were analyzed by GC-MS.
Typical procedure for adsorption of resorcinol and 1,3-cyclohexanedione. Taking the adsorption of rGO as an example. 12 mL, 8 mL and 4 mL of CH 2 Cl 2 containing 3 mg·mL −1 of resorcinol (RES) and 1,3-cyclohexanedione (1,3-CHD) were introduced to three quartz tubes (25 mL), respectively, followed by adding 50 mg of rGO into each of the tubes. The tubes were then incubated at 25 °C under sonication. Samples were taken at intervals till the adsorption reached equilibrium.
The adsorption selectivity factor of resorcinol to 1,3-cyclohexanediene was estimated as: Where q e is the equilibrium adsorption amount of the solute in the adsorbent, and c is the equilibrium concentration of solute in CH 2 Cl 2 solvent.

GC-MS analysis.
Reaction products were qualitatively measured by an Agilent 6890-5973 GC-MS instrument. The chromatographic separation was performed using a HP-5 capillary column (30.0 m × 0 .25 mm × 0.32 μ m). Helium (99.999%) was used as gas carrier with a constant flow rate of 2.0 mL·min −1 and a 1:10 split ratio. The oven was heated using the following temperature curve: initial temperature of 100 °C, heated up to 280 °C at the rate of 12 °C·min −1 and kept for 10 min. The vaporizing chamber temperature was 280 °C and the temperature of the GC/MS interface was kept at 300 °C. The mass spectrometer was operated in EI mode at 70 V of ionization voltage, 0.8 kV of detector, temperatures of ion source of 280 °C and quadrupole of 150 °C.
To quantitatively measure the reaction components, an Agilent 6820 GC equipped with HP-5 capillary column (30.0 m × 0.25 mm × 0.32 μ m) and a flame ionization detector was used. n-Dodecane was used as an internal standard.