Enhanced adsorptive removal of p-nitrophenol from water by aluminum metal–organic framework/reduced graphene oxide composite

In this study, the composite of aluminum metal–organic framework MIL-68(Al) and reduced graphene oxide (MA/RG) was synthesized via a one–step solvothermal method, and their performances for p–nitrophenol (PNP) adsorption from aqueous solution were systematically investigated. The introduction of reduced graphene oxide (RG) into MIL-68(Al) (MA) significantly changes the morphologies of the MA and increases the surface area. The MA/RG-15% prepared at RG-to-MA mass ratio of 15% shows a PNP uptake rate 64% and 123% higher than MIL-68(Al) and reduced graphene oxide (RG), respectively. The hydrogen bond and π – π dispersion were considered to be the major driving force for the spontaneous and endothermic adsorption process for PNP removal. The adsorption kinetics, which was controlled by film–diffusion and intra–particle diffusion, was greatly influenced by solution pH, ionic strength, temperature and initial PNP concentration. The adsorption kinetics and isotherms can be well delineated using pseudo–second–order and Langmuir equations, respectively. The presence of phenol or isomeric nitrophenols in the solution had minimal influence on PNP adsorption by reusable MA/RG composite.

(SEM) and transmission electron microscopy (TEM) are presented in Fig. 1. It can be seen that the MA (Fig. 1a) is in aggregative ball-liked particles and the RG (Fig. 1b) shows a flake-liked structure with wrinkles. After the RG composited with MA ( Fig. 1c-h), the MA particles are observed to attach to the surface of RG layers, and the particle size appears to be smaller. During the preparation of MA/RG, the oxygen-containing groups including -COOH and -OH on the frame of GO were sufficient coordination with the Al 3+ , which provided the dense and homogeneous nucleation nods for the formation of MIL-68(Al) with precious few agglomeration, and thus leading to the formation of nanosized and well-dispersed MIL-68(Al) crystallites 19,30 . The results of element composition (Fig. 1i) and the maps of elements (Fig. 1j-l) obtained using EDX show that the C, O and Al are uniformly distributed on the surface of the MA/RG-15% (Fig. 1h). The TEM images ( Fig. 1(m-o)) further confirm the morphological change after the hybridization. As depicted in Fig. 1o, the MA scattered in the transparent folded RG layers of MA/RG-15% composite are plate-like particles, which seemed to be different from the parental one (Fig. 1m), attributing to the distortion force of GO for MIL-68(Al) formation by the π -π stacking interaction 31 . Figure 2a shows the X-ray diffraction (XRD) patterns of MA, RG and MA/RG composites. The characteristic peaks of MA are in agreement with the previous report of MIL-68(Al) 25 , suggesting that the current material has the structure as expected. For RG, the peak at 2θ = 22.2° corresponds to graphene with the interlayer distance of 0.400 nm, indicating the reduction of graphene oxide and the restoration of sp 2 bonded carbon. The peak at 2θ = 13.2° is produced by the residual oxygen groups such as epoxy as carbonyl groups, due to incomplete reduction. After hybridization of MA with RG, the major diffraction patterns of the composites are similar to that of the pure MA, and no diffraction peak for RG is observed in the composites, which is probably due to that the low content of RG is shielded by the attached MA particles 32 . To testify the presence of RG, the Raman spectra ( Fig. 2b) of the MA/RG composites are measured. The major characteristic peaks for RG include D peak (∼ 1348 cm −1 ) which is resulted from the breathing mode of κ -point phonons of A1 g symmetry, and the G peak (∼ 1601 cm −1 ) which is from the E 2g phonon of the sp 2 hybridization 27 . For MA, the characteristic peaks at 1616, 1475 and 1147 cm −1 correspond to the in-plane vibration of the aromatic rings in the terephthalic acid ligands, and the peaks at 869 and 631 cm −1 are associated with C-H stretching or out-of-plane vibration of the aromatic rings 33 . Those peaks of MA and the D band of RG are also observed in the MA/RG composites, indicating successful hybridization of MA and RG. The MA/RG (MA/RG-1% to MA/RG-15%) present a shoulder band between 1594 and 1608 cm −1 and eventually show G peak of RG with the increase amount of RG (MA/RG-25% and MA/RG-35%), which is attributed to the integration of G band of RG with 1616 cm −1 peak of MA and the larger number of sp 2 composite aromatic rings in its structure with more incorporated RG.
The FT-IR results for the parent materials and MA/RG composites are presented in Fig. 2c. Characteristic peaks of C-H (754 cm −1 ), C-O-C (991 cm −1 ), C= O (1512 and 1701 cm −1 ) and C= C (1411 and 1587 cm −1 ) are observed for RG 34 . These peaks are also observed in the spectra of MA and MA/RG composites. Different from RG, the MA and MA/RG has the other peaks of C-OH for carboxyl (1274 cm −1 ), C-H for benzene ring (1097 cm −1 ) in terephthalic acid ligand and -OH (3425 cm -1 ) 35 . The N 2 adsorption-desorption isotherms obtained at 77 K are shown in Fig. 2d. All curves are type II isotherms typically with a type H3 hysteresis, due to the presence of mesopores 36 . It should be noted that the N 2 ad-desorption isotherms for MA/RG were above that of the MA and RG, indicating that the specific surface area of the material is increased after MA composited with RG. The MA/RG-15% has the highest specific surface area. The porous structure parameters and the pore size distributions were obtained from the N 2 adsorption data analyized using Barrete-Joynere-Halenda (BJH) model, and the results are summarized in Table 1. The MA/RG-15% exhibits the largest volume of micro and mesporous, which is consistent with the result of BET surface area measurement.
Scientific RepoRts | 6:25638 | DOI: 10.1038/srep25638 The surface element of MA/RG-15% is analyzed using the X-Ray photoelectron spectroscopy (XPS) and the results were presented in Fig. 3a. The atomic content of Al, C and O is 6.35%, 63.30% and 30.35%, respectively. The binding energy of the metal A12p is 74.92 eV (Fig. 3b), which is due to the formation of AlO 4 (OH 2 ) in the MIL-68(Al) framework. The O1s peak (Fig. 3c) at 532.40 eV accounts for the carboxylate oxygen − COO of terephthalic acid and the residue oxygen-containing groups in RG in the composites. This is further confirmed by the C1s band (Fig. 3d), which can be divided into three peaks located at 284.59, 285.55, and 289.70 eV, corresponding to the C= C/C− C, C− O and carboxylate carbon structures, respectively 37 . The p-nitrophenol (PNP) adsorption. Adsorption kinetics and rate-control mechanism. The effects of contact time on the PNP adsorption on the RG, MA and MA/RG composites are shown in Fig. 4a. For all samples, the PNP adsorption rates decrease with time until the adsorption equilibrium is reached. After hybridization of MA with RG, the adsorption of PNP are significantly enhanced, and the MA/RG-15% exhibits the best adsorption performance of 307.38 mg/g, which is 64% and 123% higher than that of MA and RG, respectively, due to the increase of surface area. The molecular size of PNP is calculated to be 0.66 nm × 0.43 nm 38 , which is smaller than the average diameter of the pores (Table 1) of MA/RG composites. Therefore, the PNP molecules can easily enter into the pore and access the surface, which favors the PNP adsorption 39 .
Scientific RepoRts | 6:25638 | DOI: 10.1038/srep25638 where q e and q t (mg/g) are the adsorption amount of PNP at equilibrium and time t(h), respectively. k 1 (h -1 ) and k 2 (g/(g.h)) are the Pseudo-firs-order and Pseudo-second-order adsorption rate constants, respectively. α (mg/(g.h)) is the initial sorption rate and β (g/mg) is related to the extent of surface coverage and activation energy for chemisorptions. Table 2 summarizes the adsorption kinetic parameters of PNP onto the tested adsorbents. Comparing with the correlation coefficients (R 2 ) of the Pseudo-first-order, Pseudo-second-order and the Elovich models, it can be concluded that the Pseudo-second-order kinetic model fits the adsorption process of all samples better than the other two. Furthermore, the deviation between calculated q e,cal and experimental q e,exp values of the Pseudo-second-order kinetic model are very lower, while that of the Pseudo-first-order kinetic model is very large. The fitting line of Pseudo-second-order is perfectly plotted in Fig. 4b, suggesting that the adsorption-determining factor of the PNP removal may be involve in the chemisorption.
To better understand the diffusion rate controlling procedure, the Intra-particle diffusion model 42 is tested as Eq. (4).
In which, i is the number of linear piecewise, k id is the Intra-particle diffusion rate constant (mg/(g.h 1/2 )), and C i is the intercept related to the thickness of the boundary layer. If the data of the whole adsorption process is good linear fit to intra-particle diffusion (i is only equal to 1) and C is zero, the intra-particle diffusion is the lonely rate limiting step, otherwise, the larger the intercept, the greater the contribution of the film diffusion sorption in the rate controlling 43 . As shown in Fig. 4c, the entire PNP adsorption process shows three linear sections in a curve, suggesting multiple steps take place during adsorption process.The piecewise fitting parameters of the Intra-particle diffusion are listed in Table 2. The values of C i for each linear portion are not zero and the correlation coefficients (R 2 ) 2 show the highest value among than (R 1 ) 2 and (R 3 ) 2 , indicating that intra-particle diffusion participate in the PNP adsorption controlling, but is not the sole rate-controlling step in all the stages, the film diffusion may also involve in the adsorption process. At the beginning of adsorption (the first segment in Fig. 4c), the film diffusion charged the mass transfer of PNP from the bulk solution to the external surface of MA/RG. As the adsorption processing (the second stage of Fig. 4c), adsorption rate starts to slow down and the intra-particle diffusion conducts the diffusion of the PNP molecules from the external surface into the pores of the MA/RG, which is also accompanied by film diffusion. In the end (the small slope section in Fig. 4c), the adsorption is reached equilibrium and the intra-particle diffusion fades out the PNP adsorption.
Adsorption isotherms and thermodynamics. The adsorption isotherm investigations are carried out under different temperature with various initial PNP concentrations. As illustrated in Fig. 5(a-c), the uptake amount of PNP increases firstly with the increase of PNP concentration and then it keeps on a horizontal (the adsorption saturation stage). The reason may be that the higher initial PNP concentration, the more strength driving force provides to overcome the mass transfer resistances when the utilization of active sites do not reach adsorption saturation. Besides, for MA (Fig. 5b) and MA/RG-15% (Fig. 5c), the PNP adsorption amount increases with increase of temperature, while for RG (Fig. 5a), the adsorbed PNP is slightly decrease as increase in temperature, which suggests that an endothermic process is for MA and MA/RG-15%, but an exothermic procedure is for RG in nature. Therefore, after MA incorporated with RG, rising the temperature favors the PNP adsorption 44 .
To gain the insight into adsorption thermodynamic behavior, the Langmuir 45 and Freundlich 18 isotherm models are used to analyze the equilibrium data according to the Eqs (5) and (6), respectively. where, C e is the equilibrium concentration (mg/L) of the PNP, q e is the amount (mg/g) of the PNP adsorbed at equilibrium and q max is the maximum adsorption capacity (mg/g), K L (L/mg) is the Langmuir constants related to energy of the adsorption. K F (L/mg) and 1/n are Freundlich constants giving an indicator of the adsorption capacity and the adsorption intensity, respectively. Table 3 lists isotherm parameters of Langmuir and Freundlich isotherms for the PNP adsorption. It can be found that, for MA, RG or MA/RG, the regression coefficients R 2 obtained from Langmuir model are much higher than that from Freundlich isotherm model, which suggests that the adsorption of PNP is best fitted with the Langmuir isotherm and the adsorption behavior is governed by monolayer adsorption on a homogenous surface. The linear relation between C e /q e and C e of Langmuir model is well ploted in Fig. 5(d-f). The maximum uptake capacity (q m,cal ) calculated from Langmuir model of the MA/RG composite at different temperature are much higher than that of the parent MA and RG. The comparisons of PNP adsorption maximum capacity with various adsorbents previously reported are listed in Table 4. It can be seen that the MA/RG composite exhibits superior PNP uptake capacity than NiAl-layered double hydroxide 1 , alumina hollow microspheres 46 , NH 2 -MIL-101(Al) 47 , carbon nanotube 3 , nanographite oxide 16 and graphene 18 , indicating that the MA/RG has great potentials for PNP removal from contaminated water. To gain insight into the essential feature of Langmuir isotherm, a dimensionless constant separation factor (R L ) is tested as the following equation 1: where, K L (L/mg) is the Langmuir constant and C 0 (mg/L) is the initial concentration of PNP in the liquid phase. The value of R L indicates whether the type of the Langmuir isotherm is unfavorable (R L > 1), linear (R L = 1), favorable (0 < R L < 1) or irreversible (R L = 0). From Table 3, the R L values for MA, RG and MA/RG are between 0 and 1, especially for MA/RG, the R L values are smaller than that for the monomers, indicating that a more favorable adsorption process is for the PNP adsorption onto the MA/RG composites. The in-depth information on inherent energetic changes related to the adsorption is provided by thermodynamic analysis. The Gibbs free energy change (∆G 0 ) of the adsorption can be calculated by the following equation 48 : where, R is the gas constant (8.314 J/mol.K), Kc is distribution coefficient (Kc = q e /C e ) at the specific temperature (K). As given in Table 5, the values of ∆G 0 are found to be negative at the tested temperature, indicating that the uptake process of PNP by RG, MA and MA/RG is spontaneous nature. By assuming that ∆H 0 and entropy change (∆S 0 ) are independent of temperature for the adsorption processes with temperatures not significantly change, the values of ∆H 0 and ∆S 0 can be evaluated by linear regression of ∆G 0 versus T as Gibbs-Helmholtz equation 49 :  As presented in Table 5, the value of ∆H 0 is positive for MA and MA/RG composite, and negative for RG, which indicates that the PNP adsorption on MA and MA/RG is an endothermic nature, but on the RG is an exothermic process. This coincides with the result of temperature effects. Furthermore, the positive (or negative) value of ∆S 0 reflects an increase (or decrease) in randomness at the solid/solution interface for MA and MA/RG (or RG) during the PNP adsorption process 50 .
Effects of pH and ionic strength. The effect of pH ranging from 3.0 to 11.0 on the adsorption capacity (q e ) of MA/ RG-15% is shown in Fig. 6a. The amount of PNP adsorbed on MA/RG-15% increases with increasing initial pH of up to 5, and then decreases with further increase in pH. The effect of pH on PNP adsorption by MA/RG is mainly resulted from the distribution of two PNP species, the molecular and the anionic. The PNP is a watersoluble solid that is moderately acidic in water (pK a = 7.15). The molecular PNP can be effectively adsorbed onto MA/RG at a lower pH by hydrophobic interaction and π -π interactions with the aromatic moieties of the PNP and the aromatic matrix of the MA/RG. However, when the pH value is below or equal to 4, a damage of MOFs starts to appear, leading to the weak binding force for PNP adsorption 42 . The PNP exists as a phenolate anion when the pH is higher than the pK a (7.15) and is hydrophilic in nature, which cannot be effectively loaded onto MA/RG composite due to the weak electron accepting ability of benzene ring in PNP for donor-acceptor interaction with MA/RG. Furthermore, the zeta potentials measurement (Fig. 6a) shows that the surface charge of MA/RG is negative when pH is greater than or equal to 9, which implies that the electrostatic repulsion between PNP anions and the MA/RG surface increases with increasing pH value, resulting in the decrease of PNP uptake amount to MA/RG adsorbent.
Effect of presence of NaCl, KCl and BaCl 2 on the PNP adsorption is depicted in Fig. 6b. It can be seen that the amount of PNP adsorbed on MA/RG is firstly decrease (or increase), and then increase (or decrease) with the increase in the concentration of NaCl, and BaCl 2 (or KCl). This depends on the strength of two different interaction mechanisms. On the one hand, the Na + , K + and Ba 2+ are trended to compete with PNP for the active site of the MA via donor-acceptor interaction, making an adverse condition for PNP adsorption. On the other hand, a favorable factor is also provided by the salt-out effect between PNP molecule and salt. This effect illustrates that solubility of non electrolyte organic compounds in solution generally displays an inverse dependency on ionic strength. Namely, in the presence of dissolved ions, water is less ordered and compressible, and the cavity volume available to accommodate PNP molecules is reduced. Thus, the solubility of PNP neutral molecule is decreased in solution, and the more decrease in solubility, the more increase is in the sorption of PNP by enhanced diffusion 51 .
Recyclability of MA/RG and PNP adsorption mechanism. The regenerability of MA/RG-15% was examined using methanol and ethanol. As seen in Fig. 6c, the PNP adsorption performance of MA/RG-15% regenerated by methanol is superior to that renewed by ethanol. After five adsorption/desorption cycles using methanol, the PNP uptake amount still remains at 271.82 mg/g, indicating that the methanol is very suitable for the regeneration of MA/RG, and the MA/RG could be a cost-effective and promising adsorbents for PNP removal. To verify the adsorption mechanisms, the FT-IR spectra (Fig. 6d) of PNP, MA/RG, mixture of MA/RG and PNP solid, and the MA/RG loaded with PNP are conducted. For PNP, the triplet bands (689, 753 and 853 cm -1 ) are the ring C-H vibrations and the bands at 1101 and 1161 cm -1 are aromatic ring vibrations; the O-H bending vibration locates at the band of 1211 cm -1 (and 3080 ~ 3500 cm -1 ) and the phenolic C-O stretch is presented at 1337 cm -1 ; the bands at 1278 and 1442 cm -1 are due to the vibration of NO 2 group; the bands at 1501 and 1594 cm -1 belong to the C-N band and aromatic stretching C= C vibrations, respectively 52 . As shown in Fig. 6d, the curve for the mixture of PNP and MA/RG (MA/RG + PNP) by mechanical blending is just a simple combination of the corresponding peaks of MA/RG and the PNP, no other obvious changes. Compare with the spectra of MA/RG + PNP, the MA/ RG loaded with PNP (MA/RG adsorbed PNP) exhibits some distinctions, indicating that many interactions are established between the PNP and MA/RG during the adsorption process. As discussed in the spectra of RG, its surface also has the epoxy ether C-O-C at the edges of broken graphene planes, which could interact with the phenolic protons to form hydrogen bond. On the basis of the disappearance of the O-H stretch for PNP (Fig. 6d) after adsorption (MA/RG adsorbed PNP), as well as the position shift of C-O-C from 991 cm -1 for RG in MA/ RG composite to 1000 cm -1 for MA/RG adsorbed with PNP, it is evident that there is hydrogen bonding between PNP and MA/RG composite 16 . The MIL-68(Al) contains numerous benzene rings with electron system, which possess the high affinity to attract the electron-withdrawing compounds by donor-acceptor interaction 40 . It is Adsorbents T(K) Kc ΔG 0 (kJ/mol) ΔS 0 (J/(K mol)) ΔH 0 (kJ/mol)  Table 5. Thermodynamic parameters for PNP adsorption.
well known that such relationship can be affected by the availability of electron density in the donor and the electron affinity of the acceptor. The nitro-substituted benzene ring can acts as electron acceptor due to the low electron density caused by strong electron withdrawing ability of NO 2 group reducing the overall electron density in the π -system of the aromatic ring 39 . As shown in Fig. 6e, the adsorption yield of nitrophenols and phenol onto MA/RG composite in case of single component is in the order of o-nitrophenol (ONP) > PNP > m-nitrophenol (MNP) ≫ phenol, indicating that the nitro substituent causes the increment in electronic acceptance of aromatic ring on nitrophenol. In fact, the π electron-rich regions in graphene layers also can interact with electron acceptor substance by π -π dispersion interaction 4 . For the multi aperture adsorbent, it is often assumed that the π − π interaction is stronger in the small pores. This agrees with the pore size distribution ( Table 1) that the larger volume of micropore, the higher adsorption capacity is for PNP onto MA/RG samples. Besides, the C= C vibrations (Fig. 6d) in aromatic ring at 1587 and 1594 cm -1 for MA/RG and PNP are located in the uniform at the band of 1600 cm -1 , indicating that the electron-rich regions in graphene layers interact with the π electron of the aromatic ring of PNP via stacking the center of the aromatic ring of the molecule on top of a graphene carbon atom and the benezene ring of the PNP on top of the graphene hexagon 18 . Owe to such force existence (Fig. 6f), it is possible to simultaneously remove isomeric nitrophenols from water. As presented in Fig. 6e, in the binary system (ONP + PNP, MNP + PNP), the MA/RG composite exhibits excellent affinity to each component with negligible change compared to that in single solution.

Conclusions
The MIL-68(Al)/reduced graphene oxide (MA/RG) composite is successfully synthesized via a simple solvothermal method. The presence of reduced graphene oxide (RG) changes the morphology and surface area of composites, but not the crystalline structure. The surface area of composite is firstly increase and then decrease with increasing RG, and the PNP adsorption capacity exhibits the same trend with the maximum uptakes calculated from Langmuir model of 332.23 mg/g for MA/RG-15% at 303 K, which is much super than the MA and RG individual. This good performance is linked to improvement of porosity, the hydrogen bond and π -π dispersion interaction between PNP and the composite. The solution pH, ionic strength, temperature and initial PNP concentration extremely affect the PNP adsorption, but the presence of phenol and isomerism nitrophenols has a slightly influence on PNP removal. The adsorption process involved in film-diffusion and intra-particle diffusion beys well with the Pseudo-second-order model and the Langmuir model. The coupling of MOFs with reduced graphene oxide provides a favorable pathway to synthesis high reusability and effective adsorbent forsimultaneous removal of nitrophenols from wastewater.

Methods
Materials. Graphite powder (particle size < 30 um) was purchased from Tianjin Kermel Chemical Regent Scientific RepoRts | 6:25638 | DOI: 10.1038/srep25638 the supernatant was measured using UV spectrophotometer (UV-2550, SHIMADZU, Japan) at 270, 278, 273 and 317 nm, respectively. The adsorptive quantity was calculated as follows: where q t (mg/g) is the adsorption quantity; C 0 and C t are the pollution concentration of initial and interval time (mg/L), respectively; V(L) is the volume of solution, and m (g) is the weight of adsorbent.