Materials Chemistry and Physics 148 (2014) 153e158
Contents lists available at ScienceDirect
Materials Chemistry and Physics journal homepage: www.elsevier.com/locate/matchemphys
Mesoporous hydroxyapatite: Preparation, drug adsorption, and release properties Lina Gu, Xiaomei He, Zhenyu Wu* School of Chemistry & Chemical Engineering, Anhui University, Hefei 230039, China
h i g h l i g h t s Mesoporous HA was synthesized by a simple precipitation method without any template. The kinetics of adsorption followed the pseudo-second-order rate expression. Thermodynamics investigation showed that adsorption was spontaneous and endothermic. DOX-loaded HA showed a long-term, steady, and pH-controlled release rate.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 20 March 2014 Received in revised form 25 June 2014 Accepted 20 July 2014 Available online 6 August 2014
Mesoporous hydroxyapatite (HA) was synthesized through gaseliquid chemical precipitation method at ambient temperature without any template. Structure, morphology and pore size distribution of HA were analyzed via X-ray powder diffraction, scanning electron microscopy, transmission electron microscopy, high-resolution electron microscopy and N2 adsorption/desorption. The chemotherapeutic agent doxorubicin (DOX) was used to investigate the drug adsorption and release behavior of HA. The kinetics of DOX adsorption on HA followed the pseudo-second-order rate expression. Adsorption isotherms at various temperatures were obtained, and the equilibrium data ﬁtted the Langmuir model. The values of thermodynamic parameters (Gibbs free energy, entropy, and enthalpy changes) demonstrated that the adsorption process was spontaneous and endothermic. In vitro pH-responsive (pH ¼ 7.4, 5.8) controlled release was investigated. DOX-loaded HA showed a slow, long-term, and steady release rate. The release rate at pH5.8 was larger than that at pH7.4. Consequently, the as-prepared mesoporous HA has potential applications in controlled drug delivery systems. © 2014 Elsevier B.V. All rights reserved.
Keywords: Surfaces Precipitation Adsorption Desorption
1. Introduction Hydroxyapatite [HA; Ca10 (PO4)6(OH)2] is a chemical analog of the bone tissue mineral component. In bones and teeth, HA is present in the needle-like nanocrystalline state and is embedded in the collagen matrix . So, HA exhibits excellent bioactivity, biocompatibility, and high compressive strength . Due to these properties, HA (mesoporous particles, hollow microspheres, etc.) has been widely applied to controlled delivery systems for proteins [3e5], drugs [6e10], and genes  in the past few decades. Piskounova and her colleagues had shown that HA was an excellent tool for delivery of the bone morphogenetic protein BMP-2 both in vitro and in vivo [12,13]. Brohede investigated the fast-loading
* Corresponding author. Tel.: þ86 551 63861326; fax: þ86 551 63861279. E-mail address: [email protected]
(Z. Wu). http://dx.doi.org/10.1016/j.matchemphys.2014.07.024 0254-0584/© 2014 Elsevier B.V. All rights reserved.
slow-release biomimetic hydroxyapatite coatings on surgical implant with the antibiotics amoxicillin, gentamicin sulfate, tobramycin and cephalothin . Also bisphosphonate delivery had been extensively investigated . Forsgren successfully incorporated bisphosphonates and antibiotics (cephalothin) simultaneously into a biomimetic HA implant coating . Chen used arginine-modiﬁed HA nanoparticles for DNA enzymes delivery which was therapeutic applied in a nasopharyngeal carcinoma model . The goal of controlled drug delivery systems is to achieve a constant, controlled, and long-period release rate. However, obvious initial burst release and short-term release were often observed. Mizushima  suggested that modifying the manufacturing method of HA and adding other substances were useful to delay drug release. Wang  prepared Poly (lactide-coglycolide) coated HA microspheres which showed signiﬁcant slower drug release and lower initial burst than that of HA
L. Gu et al. / Materials Chemistry and Physics 148 (2014) 153e158
Ca(NO3)2$4H2O solution was adjusted to 10.5 ~ 11 using continuously generated NH3 gas. Under vigorous stirring, the aqueous solution of (NH4)2HPO4 was added dropwise to the Ca(NO3)2$4H2O solution, and the resulting precipitates were stirred for another 2 h and stored for 24 h. The aged precipitate was washed by water and anhydrous ethanol and dried at 60 C for 24 h.
microspheres. Li  synthesized alendronate functionalized HA, and the materials showed relatively slower release rate compared with HA. Recently, Steckel and his team  investigated that HA coatings deposited on TiO2 coated ﬁxation pins could exhibit bactericidal effects against staphylococcus aureus in agar medium for 6 days after loading with the antibiotics tobramycin for only 5 min. The same team also very recently showed that by optimizing the HA deposition process the tobramycin delivering HA coatings could deliver pharmaceutically relevant amounts of tobramycin over 12 days . It is well known that biological performance of drug delivery systems are controlled by the drug adsorption/desorption (release) behaviors which depend on the structure, properties and the surrounded environment [23,24]. Compared to HA nanoparticles, mesoporous HA exhibited higher drugeloading capacity and enhanced drug release efﬁcacy, which was due to its large surface areas and high pore volumes . In this paper, we present a simple, modiﬁed precipitation method to synthesize mesoporous HA. Pure HA was obtained at ambient temperature without any template. Using a water-soluble anticancer drug, doxorubicin (DOX), as a drug model, the adsorption kinetics and thermodynamics of DOX on HA were investigated. Release proﬁle exhibited that DOX-loaded HA had a slow, sustained, and pH-controlled release rate in the phosphate buffered saline (PBS).
Phase analysis was conducted via Purkinje XD-3 powder X-ray diffraction (XRD) with a Rigaku D/max-gA rotation anode X-ray diffractometer (CuKa, l ¼ 0.15418 nm). To examine the size and morphology of the as-synthesized samples, transmission electron microscopy (TEM) and high resolution transmission electron microscopy (HRTEM) images were obtained using a JEOL JEM-2100 high resolution electron microscopy with an accelerating voltage of 200 kV. Scanning electron microscopy (SEM) images were obtained using a HITACHI S4800 scanning electron microscope operated at 5 kV. The speciﬁc surface area and pore size distribution were determined using MICROMERITICS ASAP2020M þ C surface area and porosity analyzer with a degas temperature of 105 C and an outgas time of 12 h according to the BrunauereEmmetteTeller (BET) equation and the method of BarretteJoynereHalenda (BJH), respectively.
2. Materials and methods
2.4. Adsorption of DOX
DOX solutions in various concentrations were freshly prepared. To conﬁrm the time needed for adsorption equilibrium establishment, adsorption kinetics was investigated. HA at 10 mg was mixed with 5 ml DOX solution (734 mg ml1). The resulting suspension was shaken (220r/min) in a water bath shaker (HQ45B, Wuhan Science and Technology Instrument Factory) at 301 K for 24 h. At certain time intervals, samples were withdrawn from the suspension and centrifuged. The concentrations of DOX in the supernatants, ct, were then determined through ultravioletevisible (UVevis) spectroscopy (Purkinje TU-1901) at a ﬁxed wavelength of 480 nm. The adsorption quality of DOX was deﬁned as follows:
Calcium nitrate (Ca(NO3)2$4H2O), ammonium phosphate ((NH4)2HPO4) and ammonia solution (NH3$H2O) were purchased from Tianjin Guangfu Fine Chemical Research Institute (China). Anhydrous ethanol was obtained from Shanghai Zhenxing First Chemical Industry Factory (China). DOX was purchased from Yongnuo Pharmaceutical Factory. All reagents were of analytical purity and used without further puriﬁcation. 2.2. Preparation of HA HA was prepared via a gaseliquid chemical precipitation method. Ethanol solution of Ca(NO3)2$4H2O (12.5 mmol, 50 ml) and aqueous solution of (NH4)2HPO4 (7.5 mmol, 50 ml) with Ca/P ratio of 1.67 were prepared. As shown in Fig. 1, the ethanol solution of Ca(NO3)2$4H2O was continuously stirred and placed in a bigger sealed beaker with the ammonia solution. The pH of
ðco ct Þ$V m
where Qt is the amount of DOX adsorbed on HA (mg mg1), co and ct are the initial and residual concentrations at time t of DOX, V is the volume of the DOX solution, and m is the mass of HA (mg). For the determination of the maximum amount of DOX, 10 mg HA was impregnated with 5 ml DOX aqueous solution at different concentrations. The suspensions were shaken for 22 h at 296, 301 K, and 306 K. After equilibrium, the liquid and solid phases were separated by centrifugation, and the residual drug concentrations in the liquid phase were determined. 2.5. In vitro release of DOX
Fig. 1. Schematic representation of the experimental setting.
HA (0.1 g) was immersed in 40 mg ml1 DOX solution and shaken for 22 h. The powders were isolated through ﬁltration, washed with a small volume of water, and dried vacuum-dried to obtain the DOX-loaded HA. In vitro release of DOX from HA was performed at 37 C in PBS (pH ¼ 7.4, 5.8). DOX-loaded HA (84.57 mg DOX/mg HA) at 25 mg was immersed in the release medium (10 ml). At predetermined time intervals, 6 ml buffer solution was obtained for UVevis analysis, after which the same volume of fresh buffer solution was injected into the system.
L. Gu et al. / Materials Chemistry and Physics 148 (2014) 153e158
Fig. 2. XRD pattern of the as-prepared HA at room temperature. Fig. 4. N2 adsorptionedesorption isotherm and BJH pore size distribution curve (inset) of HA.
3. Results and discussion 3.1. Characterization of the as-prepared HA The XRD pattern of the as-prepared powders is shown in Fig. 2. The phase composition of the powders was determined to be hexagonal HA according to the standard JCPDS card (No. 09-0432). No impurity phase was detected. HA prepared using this method has high purity. The synthesis was performed at room temperature, but the strong diffraction peaks indicated that the as-prepared HA was well crystallized. SEM and TEM micrographs of the as-prepared HA are shown in Fig. 3. HA particles were well dispersed and resembled uniform spindles that were 20 nm in width and 100 nme150 nm in length
(Fig. 3a and b). In Fig. 3c, a typical TEM image of a single HA particle clearly revealed that the spindle-shaped HA particle comprised homogenous, small nanocrystalline particles (approximately 6 nm). Mesopores (2 nme10 nm) were observed among these smaller particles. The crystalline structure of mesoporous HA was further conﬁrmed by HRTEM and shown in Fig. 3d. Clear lattice fringes with interplanar spacing of 0.47 nm corresponding to the (1 1 0) plane of hexagonal HA were also observed. Nitrogen adsorptionedesorption isotherm and BJH pore size distribution curve (inset) of HA are shown in Fig. 4. The nitrogen adsorptionedesorption isotherm showed type IV isotherm behavior with a distinct hysteresis loop indicating that the asprepared HA was mainly of a mesopore in the size range of
Fig. 3. Images of the as-prepared HA powders (a) SEM, (b) and (c) TEM (d) HRTEM.
L. Gu et al. / Materials Chemistry and Physics 148 (2014) 153e158
2 nme50 nm . The centers of mesopores were 2.3, 3.6, and 32 nm according to the BJH pore size distribution analysis. Most pores were less than 6 nm in size, which agreed well with the TEM results. The pores centered at 32 nm were probably voids among the spindle-shaped HA particles. In addition, the total pore volume of the HA was 0.508 ml g1, and the BET surface area was calculated to be 116.8 m2 g1; these values were larger than those of hollow HA microspheres  and mesoporous HA nanoparticles prepared through hydrothermal methods . The as-prepared HA has high surface area and porosity, indicating that the HA prepared in this study has potential applications in drug delivery, as well as water treatment and puriﬁcation systems. 3.2. Adsorption kinetics of DOX on HA The adsorption amounts of DOX from aqueous solution to the HA as a function of the adsorption time are shown in Fig. 5a. The amount of DOX adsorbed on HA increased with increasing contact time, until adsorption equilibrium was established within 22 h. Contrary to other models, the pseudo-second-order equation is based on the sorption capacity of the solid phase and can predict behavior over a wide range of studies . Therefore, Ho pseudosecond-order rate equation  was used to describe the adsorption kinetics of DOX to HA as follows:
t 1 t ¼ þ Q t k2 Q 2e Q e
where Qt (mg mg1) is the absorption amount at time t, Qe (mg mg1) is the equilibrium absorption capacity, and k2 is the rate constant of pseudo-second-order kinetics. Fig. 5b shows good linearization of t/ Qt versus t in the entire adsorption process. The correlation coefﬁcient (R2 ¼ 0.987) was close to 1, suggesting that the adsorption of DOX on HA is a pseudo-second-order kinetic process. To analyze the diffusion mechanism, an intra-particle diffusion model was used with the following equation:
Q t ¼ ki $t 1=2
where ki is the intra-particle diffusion rate constant, mg mg1 h1/2. Nassar  stated that the adsorption process was controlled by the rate of intra-particle diffusion when the adsorbed quantity varied almost proportionately with t1/2 rather than t. Fig. 5c shows that plotting the adsorbed amount against the square root of time resulted in satisfactory linearization of the entire adsorption process (R2 ¼ 0.997). However, the line did not pass the origin; thus intra-particle diffusion was a rate-limiting step, but it was not the only one . The adsorption rate was also controlled by other inter-particle diffusion processes, such as surface adsorption and liquid ﬁlm diffusion.
Fig. 6. Adsorption isotherms of DOX on HA at three different temperatures.
3.3. Adsorption thermodynamics of DOX on HA The adsorption isotherm curves at three different temperatures (296, 301, and 306 K) are presented in Fig. 6. At lower initial concentration of DOX, the adsorption isotherm curves were almost linear and the adsorbed amounts increased quickly. The curves progressively bent and exhibited a tendency to become ﬂat. The adsorption process increased with increasing temperature, indicating that the adsorption of DOX on HA was satisfactory at high temperatures. To describe the adsorption equilibrium, the relationship between the adsorbed amount (Qe, mg mg1) and its equilibrium concentration (ce, mg ml1) was simulated using two common models, namely, Langmuir and Freundlich. Table 1 shows the ﬁt results simulated from the two models. The correlation coefﬁcients (R2) for Langmuir and Freundlich isotherms indicated that the Langmuir model ﬁt the experimental data better than the Freundlich model. Moreover, adsorption was shown to increase exponentially in the Freundlich model . HA had a limited adsorption capacity. Thus, the Langmuir model is a better ﬁt for the adsorption process. Thermodynamic parameters (the Gibbs free energy change, DG0; entropy change, DS0; and enthalpy change, DH0) were calculated from the adsorption isotherms. The value of DG0 was calculated using the following equation:
DG0 ¼ RT ln K 0
Fig. 5. Plots of (a) Qt versus t, (b) t/Qt versus t and (c) Qt versus t1/2 for adsorption of DOX on HA.
L. Gu et al. / Materials Chemistry and Physics 148 (2014) 153e158 Table 1 Isotherm's equations and simulated results. T (K)
Langmuir 1 Qe
296 301 306
Table 2 Constants of linear ﬁt of lnKd versus ce (lnKd ¼ a þ b ce) for DOX adsorption on HA.
¼ Qm1$Kc $c1e þ Q1m
ln Qe ¼ ln Kf þ 1n ln ce
Qm (mg mg1)
Kc (ml g1)
236.407 316.456 270.270
0.8098 0.6556 0.4305
0.979 0.994 0.987
0512 0.631 0.117
1.282 1.303 0.962
0.972 0.974 0.973
296 301 306
5.274 5.340 5.359
5.946 4.959 3.845
0.965 0.983 0.966
Table 3 Values of thermodynamic parameters for DOX adsorption on HA.
where K0 is the equilibrium constant. The values of lnK0 are obtained by plotting lnKd versus ce (Fig. 7) and extrapolating ce to zero. Kd is the distribution coefﬁcient and can be calculated using the following equation:
c0 ce V $ m ce
where m (mg) is the mass of HA and V (ml) is the volume of the solution. The values of the entropy change DS0 and the enthalpy change DH0 were calculated according to the van't Hoff equation as follows:
ln K 0 ¼
DS0 DH0 1 $ R R T
DG0 (KJ mol1)
DS0 (J/K mol)
DH0 (KJ mol1)
296 301 306
195.2 208.5 212.5
12.98 13.36 13.63
process. In liquidesolid adsorption, the solute and solvent were adsorbed simultaneously on the surface of the solid. The entropy change depended on the interaction forces between the solute/ solvent and the solid surface, as well as the molecule volumes of solute and solvent. Hydrogen bonds between DOX and HA and the
Constants of the linear ﬁt of lnKd versus ce are listed in Table 2. The values derived from Eqs. (4) and (6) are tabulated in Table 3. The negative values of DG0 showed that adsorption of DOX on HA was spontaneous. The value of DG0 decreased with increasing temperature, indicating that the adsorption increased at high temperature. The values of the entropy change DS0 and the enthalpy change DH0 simulated from the linear ﬁt of lnK0 versus 1/T were positive, demonstrating that the adsorption of DOX on HA was an endothermic and increasing entropy process. The low value of enthalpy change DH0 indicated that adsorption of DOX on HA should be considered as physical adsorption. To conﬁrm this assumption, the UVevis diffused reﬂection spectrum was used. As shown in Fig. 8, the UVevis spectra of DOX and DOXloaded HA were similar, and no extra peaks appeared when DOX was adsorbed on HA. Therefore, the adsorption force between DOX and HA was physical. Moreover, the value of entropy change DS0 was positive, indicating that adsorption was an entropy-driven
Fig. 8. UVevis spectra of (a) HA, (b) DOX and (c) DOX-loaded HA.
Fig. 7. Linear plots of lnKd versus ce of DOX adsorption on HA.
Fig. 9. In vitro release curves of DOX-loaded HA in PBS (pH ¼ 7.4, 5.8).
L. Gu et al. / Materials Chemistry and Physics 148 (2014) 153e158
high molecular volume of DOX indicated that DOX occupied more adsorption sites than the water molecule. When a DOX molecule was adsorbed on HA, water molecules were displaced. The entropy change of water desorption from HA (increasing entropy) was larger than that of DOX adsorption on HA (decreasing entropy). Thus, the total adsorption entropy change was positive. 3.4. Drug release of DOX-loaded HA Given that the physiological pH in the blood stream is 7.4 and the pH value in endosomes is in the range of 5.5e6.4 , drugrelease experiments were performed in PBS (pH ¼ 7.4, 5.8) at 37 C. As shown in Fig. 9, rapid release occurred within 9 h, followed by a relatively slow release until 120 h. After the initial rapid release, the released DOX concentrations in PBS were maintained at approximately 1.24 and 2.31 mg ml1 under pH 7.4 and 5.8 respectively. The released concentration of DOX from HA at pH 5.8 was higher than that at pH 7.4. This ﬁnding can be explained by the following reasons. First, DOX on the surface of HA could obtain protons and change from the amino group (eNH2) to the tertiary amine (eNHþ 3 ) in acidic medium. Thus, the hydrogen bonds between DOX and HA cannot form . Second, the protonated DOX has high solubility . As a result, the release rate of DOX increases with decreasing pH value of the buffer solution. Furthermore, we found that the released DOX concentrations in PBS solution had a limiting value. The concentration of DOX did not increase even at prolonged withdrawal time interval. This fact indicated that the DOX-loaded HA had a slow, long-term, and steady release rate, which could prevent the explosive release of the drug and prolong the drug effect. Therefore, the as-prepared HA could be a viable candidate for clinical application, as a sustained long-term drug concentration is secured. 4. Conclusions Mesoporous HA was prepared through gaseliquid chemical precipitation method using with NH3 as a pH-adjusting agent. TEM and BET measurements demonstrated that the as-prepared HA nanoparticles were well-dispersed and mesoporous. The adsorption kinetics and thermodynamics of DOX on HA were investigated, and results showed that the adsorption was spontaneous and endothermic. DOX-loaded HA exhibited a slow, long-term and steady release rate in PBS. The release rate can be adjusted by varying the pH value of the release medium. Therefore, HA prepared in this study can be a good candidate for sustained drug release. Acknowledgments The authors are grateful for the ﬁnancial support by the National Natural Science Foundation of China (No. 21001001) and the Doctoral Research Foundation of Anhui University (No.0230331933190097; 33190184).
References  Y. Wang, S. Zhang, K. Wei, N. Zhao, J. Chen, X. Wang, Mater. Lett. 60 (2006) 1484e1487.  S. Higashi, T. Yamamuro, T. Nakamura, Y. Ikada, S.H. Hyon, K. Jamshidi, Biomaterials 7 (1986) 183e187.  S.K. Swain, D. Sarkar, Appl. Surf. Sci. 286 (2013) 99e103.  H. Fu, M.N. Rahaman, D.E. Day, R.F. Brown, J. Mater. Sci. 22 (2011) 579e591.  K. Tomoda, H. Ariizumi, T. Nakaji, K. Makino, Colloids Surf. B 76 (2010) 226e235.  K.L. Lin, Y.L. Zhou, Y. Zhou, H.Y. Qu, F. Chen, Y.J. Zhu, J. Chang, J. Mater. Chem. 21 (2011) 16558e16565.  F. Ye, H.F. Guo, H.J. Zhang, X.L. He, Acta Biomater. 6 (2010) 2212e2218.  C.A.S. Souza, A.P.V. Colombo, R.M. Souto, C.M. Silva-Boghossian, J.M. Granjeiro, ~o, Colloids Surf. B 87 (2011) 310e318. G.G. Alves, A.M. Rossi, M.H.M. Rocha-Lea  Y.P. Guo, Y.B. Yao, Y.J. Guo, C.Q. Ning, Microporous Mesoporous Mater. 155 (2012) 245e251. o sarczyk, J. Szymura-Oleksiak, B. Mycek, Biomaterials 21 (2000)  A. Sl 1215e1221.  S.R. Bhattarai, S. Aryal, K.C.R. Bahadur, N. Bhattarai, P.H. Hwang, H.K. Yi, H.K. Kim, Mater. Sci. Eng. C 28 (2007) 64e69.  S. Piskounova, J. Forsgren, U. Brohede, H. Engqvist, M. Strømme, J. Biomed. Mater. Res.: Part B 91B (2) (2009) 780e787.  J. Forsgren, U. Brohede, S. Piskounova, A. Mihranyan, S. Larsson, M. Strømme, H. Engqvist, J. Biomater. Nanobiotechnol. 2 (2011) 150e155.  U. Brohede, J. Forsgren, S. Roos, A. Mihranyan, H. Engqvist, M. Strømme, J. Mater. Sci. Mater. Med. 20 (2009) 1859e1867.  J. Åberg, U. Brohede, A. . Mihranyan, M. Strømme, H. Engqvist, J. Mater. Sci. Mater. Med. 20 (2009) 2053e2061.  J. Forsgren, U. Brohede, M. Strømme, H. Engqvist, Biotechnol. Lett. 33 (2011) 1265e1268.  Y. Chen, L.F. Yang, S.P. Huang, Z. Li, L. Zhang, J. He, Z.J. Xu, L.Y. Liu, Y. Cao, L.Q. Sun, Int. J. Nanomed. 8 (2013) 3107e3118.  Y. Mizushima, T. Ikoma, J. Tanaka, K. Hoshi, T. Ishihara, Y. Ogawa, A. Ueno, J. Controlled Release 110 (2006) 260e265.  S.N. Wang, X.Y. Wang, H. Xu, H. Abe, Z.Q. Tan, Y.Q. Zhao, J.B. Guo, M. Naito, H. Ichikawa, Y. Fukumori, Adv. Powder Technol. 21 (2010) 268e272.  D.D. Li, Y.T. Zhu, Z.Q. Liang, Mater. Res. Bull. 48 (2013) 2201e2204. €rensen, M. Lilja, T.C. So €rensen, M. Åstrand, P. Procter, S. Fuchs,  J.H. So M. Strømme, H. Steckel, J. Biomed. Mater. Res. Part B (2014), http://dx.doi.org/ 10.1002/jbm.b.33117. €rensen, M. Lilja, M. Åstrand, T.C. So €rensen, P. Procter, M. Strømme,  J.H. So H. Steckel, Curr. Drug Delivery 11 (2014) 501e510.  N. Ribeiro, S.R. Sousa, F.J. Monteiro, J. Colloid Interface Sci. 351 (2010) 398e406.  A. Generosi, V.V. Smirnov, J.V. Rau, V. Rossi Albertini, D. Ferro, S.M. Barinov, Mater. Res. Bull. 43 (2008) 561e571.  Y.H. Yang, C.H. Liu, Y.H. Liang, F.H. Lin, K.C.W. Wu, J. Mater. Chem. B 1 (2013) 2447e2450.  M. Kruk, M. Jaroniec, Chem. Mater. 13 (2001) 3169e3183.  Y. Jiao, Y.P. Lu, G.Y. Xiao, W.H. Xu, R.F. Zhu, Powder Technol. 217 (2011) 581e584.  Q.F. Zhao, T.Y. Wang, J. Wang, L. Zheng, T.Y. Jiang, G. Cheng, S.L. Wang, Appl. Surf. Sci. 257 (2011) 10126e10133.  Y.S. Ho, G. Mckay, Trans. IChemE 76 (1998) 332e340.  Y.S. Ho, D.A.J. Wase, C.F. Forster, Environ. Technol. 17 (1996) 71e77.  M.M. Nassar, Water Sci. Technol. 40 (1999) 133e139.  W.J. Weber Jr., J.C. Morris, J. Sanit. Eng. Div., Am. Soc. Civ. Eng. 89 (1963) 31e59.  C.L. Chen, X.L. Li, D.L. Zhao, X.L. Tan, X.K. Wang, Colloids Surf. A 302 (2007) 449e454.  M. Guo, Y. Yan, X. Liu, H. Yan, K. Liu, H. Zhang, Y. Cao, Nanoscale 2 (2010) 434e441.  X.Y. Yang, L. Chen, B. Han, X.L. Yang, H.Q. Duan, Polymer 51 (2010) 2533e2539.  J.N. Qi, P. Yao, F. He, C.L. Yu, C. Huang, Int. J. Pharm. 393 (2010) 177e185.