Journal of Hazardous Materials 254–255 (2013) 345–353 Contents lists available at SciVerse ScienceDirect Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat Adsorption of pharmaceuticals onto trimethylsilylated mesoporous SBA-15 Tung Xuan Bui a , Viet Hung Pham b , Son Thanh Le c , Heechul Choi a,∗ a School of Environmental Science and Engineering, Gwangju Institute of Science and Technology (GIST), 261 Cheomdan-gwagiro, Buk-gu, Gwangju 500-712, Republic of Korea b Center of Environment and Sustainable Development, College of Science, Vietnam National University, 334 Nguyen Trai, Thanh Xuan, Hanoi, Vietnam c Faculty of Chemistry, College of Science, Vietnam National University, 19 Le Thanh Tong, Hoan Kiem, Hanoi, Vietnam h i g h l i g h t s • • • • • Trimethylsilylated SBA-15 for adsorptive removal of a mixture of 12 pharmaceuticals Hydrophobic interaction as a primary driving force in the adsorption The rate-limiting steps were diffusion through boundary layer, mesopores and micropores Adsorption efficiency was not changed as pharmaceutical concentration varied Adsorption efficiency was not influenced by the presence of natural organic matters a r t i c l e i n f o Article history: Received 10 October 2012 Received in revised form April 2013 Accepted April 2013 Available online 10 April 2013 Keywords: Adsorption Trimethylsilylated mesoporous SBA-15 Pharmaceuticals Kinetics Natural organic matter a b s t r a c t The adsorption of a complex mixture of 12 selected pharmaceuticals to trimethylsilylated mesoporous SBA-15 (TMS-SBA-15) has been investigated by batch adsorption experiments The adsorption of pharmaceuticals to TMS-SBA-15 was highly dependent on the solution pH and pharmaceutical properties (i.e., hydrophobicity (log Kow ) and acidity (pKa )) Good log–log linear relationships between the adsorption (Kd ) pH and pH-dependent octanol–water coefficients (Kow ) were then established among the neutral, anionic, and cationic compounds, suggesting hydrophobic interaction as a primary driving force in the adsorption In addition, the neutral species of each compound accounted for a major contribution to the overall compound adsorption onto TMS-SBA-15 The adsorption kinetics of pharmaceuticals was evaluated by the nonlinear first-order and pseudo-second-order models The first-order model gave a better fit for five pharmaceuticals with lower adsorption capacity, whereas the pseudo-second-order model fitted better for seven pharmaceuticals having higher adsorption capacity In the same group of properties, pharmaceuticals having higher adsorption capacity exhibited faster adsorption rates The rate-limiting steps for adsorption of pharmaceuticals onto TMS-SBA-15 are boundary layer diffusion and intraparticle diffusion including diffusion in mesopores and micropores In addition, the adsorption of pharmaceuticals to TMSSBA-15 was not influenced by the change of initial pharmaceutical concentration (10–100 g L−1 ) and the presence of natural organic matter © 2013 Elsevier B.V All rights reserved Introduction Recently, pharmaceuticals have been considered as emerging micropollutants [1,2] and then have been received a lot of attention from scientists [3–5] Numerous pharmaceuticals have been frequently observed up to g L−1 level in aqueous environment [4–7] and as high as 100 g L−1 in effluents from drug manufacturers [4,8] The ubiquitous occurrence of pharmaceuticals in the environment is attributable to the fact that the technologies used in current ∗ Corresponding author Tel.: +82 62 715 2441; fax: +82 62 715 2434 E-mail address: hcchoi@gist.ac.kr (H Choi) 0304-3894/$ – see front matter © 2013 Elsevier B.V All rights reserved http://dx.doi.org/10.1016/j.jhazmat.2013.04.003 wastewater treatment systems are not designed for the removal of pharmaceuticals and therefore are not sufficient for eliminating many pharmaceuticals [9] Therefore, alternative treatment technologies that achieve efficient pharmaceutical removal need to be developed [2] Among several technologies proposed for the removal of pharmaceuticals from water, for example, nanofiltration [10], advanced oxidation processes [11], and adsorption [12], adsorption has been receiving special attention owing to its simplicity, low setup and operation cost, and no production of undesirable byproducts Several studies have evaluated the adsorption of pharmaceuticals on zeolites [12], activated carbon [13], montmorillonite [14], and mesoporous silica [15] Mesoporous materials 346 T.X Bui et al / Journal of Hazardous Materials 254–255 (2013) 345–353 with high surface area, large and uniform pore size, high pore volume, and tailorable surface can be good candidates for the adsorptive removal of pharmaceuticals Mesoporous silica SBA-15 was reported to have high adsorption efficiency for several neutral and acidic pharmaceuticals in acidic conditions [15] However, the adsorption efficiency of SBA-15 was significantly reduced at a neutral pH, especially for acidic pharmaceuticals, due to the hydrophilic and negatively charged nature of SBA-15 surface in the media The results suggest that a modification of SBA-15 surface must be conducted to achieve a better efficiency Mesoporous silica SBA-15 grafted with copper(II) amine complex [16] showed high adsorption capacity toward naproxen at alkaline conditions Recently, we have found that the attachment of hydrophobic trimethylsilyl group on SBA-15 resulted in a significant improvement in the adsorption efficiency for a series of pharmaceuticals in comparison with bare SBA-15 and other organic group grafted SBA-15 materials as well [17] The number of pharmaceuticals achieving high adsorption efficiency onto trimethylsilylated SBA-15 did not alter significantly as the pH was varied in the range of 5.5–7.6 [17] However, the previous study mainly focused on qualitatively interpreting pharmaceutical adsorption behavior on trimethylsilylated SBA-15 in different pH media and explaining the difference in adsorption efficiency between trimethylsilylated SBA-15 and SBA-15 To evaluate comprehensively the effectiveness and applicability of trimethylsilylated SBA-15, it is indispensable to implement further studies, such as: (1) adsorption kinetics, (2) effect of adsorbate concentration, (3) impact of adsorbent amount, and (4) interference of organic substances (e.g natural organic matter (NOM)) to the adsorption process These issues are all addressed in this study; in addition, this study aims to quantify the correlation between adsorbate properties and adsorption efficiency and to explore the adsorption driving force, pharmaceutical species as major contributor to the adsorption, and the rate-limiting step of the adsorption process These studies shall not only gain insight into the adsorption but also provide knowledge and tools to control the adsorption process and to implement computation for a pilot design in the future Materials and methods 2.1 Chemicals and adsorbates Reagent grade chemicals used in this study included NaCl (99%, Duksan), NaOH (98%, DC chemicals), 2(N-morpholino)ethanesulfonic acid (MES 99%, Sigma), 3-(N-morpholino)propanesulfonic acid (MOPS 99.5%, Sigma), NaN3 (99.5%, Sigma–Aldrich), tetraethylorthosilicate (TEOS 98%, Aldrich), Pluronic P123 (BASF), hydrochloric acid (HCl, 37%), hexamethyldisilazane (HMDS 99%, Aldrich), anhydrous toluene (99.8%, Sigma–Aldrich), absolute ethanol (HPLC grade, Fisher scientific), methanol (HPLC grade, Fisher scientific), and acetone (HPLC grade, Fisher scientific) Suwannee River NOM (catalog no 1R101N), Suwannee River fulvic acid (FA) (catalog no 2S101F), and Suwannee River humic acid (HA) (catalog no 2S101H) were purchased from the International Society of Humic Substances (ISHS) and used as received; the analytical data of these NOM can be found elsewhere [18] Twelve selected pharmaceuticals for testing and nine surrogate standards for analysis were purchased from Sigma–Aldrich, United States Pharmacopeia, and Toronto Research Chemicals The list of studied pharmaceuticals together with their pKa and Kow are described in Table Other related information and physicochemical properties of pharmaceuticals and surrogates can be found elsewhere [17] 2.2 Preparation of trimethylsilylated SBA-15 The preparation of trimethylsilylated SBA-15 (TMS-SBA-15) was adapted from literature [19–21] First, SBA-15 was synthesized by a method described elsewhere [15,22] Calcined SBA-15 (1.0 g) was heated at 150 ◦ C in vacuo for h and then slurried in anhydrous toluene (100 mL) for h Next, a five-fold excess amount of HMDS (3.0 mL) in anhydrous toluene (50 mL) was added and stirring was continued at room temperature [21] for 24 h The resulting product was filtered and washed with different solvents in the following order: toluene, acetone, ethanol, ethanol/water (50:50, v/v), water, ethanol, and acetone The sample was then dried in vacuo at 150 ◦ C for h and stored in a desiccator 2.3 Characterization The synthesized TMS-SBA-15 material was characterized by X-ray powder diffraction (XRD) using a PANalytical X’Pert PRO˚ Nitrogen MPD diffractometer with Cu K␣ radiation ( = 1.5406 A) adsorption–desorption measurement was carried out using a Micromeritics ASAP 2020 analyzer at −196 ◦ C after the sample was degassed at 100 ◦ C for 12 h Fourier transform infrared (FTIR) spectra were recorded on a JASCO FTIR-460 Plus spectrometer via the KBr pellet method CHN elemental analysis was conducted on an EA-1110 (Thermoquest) elemental analyzer Detail characterization of the material can be found in the previous report [17] 2.4 Batch adsorption and desorption experiment Batch adsorption experiments were studied in 40-mL amber vials in 10 mL of aqueous solutions containing 100 mg L−1 of sodium azide and enough amount of NaCl to ensure a total ionic strength of 10 mM The solution pH was controlled using either a mM buffer of MES (pH 5.5) or MOPS (pH 6.6 and 7.6) MES and MOPS were reported to have a negligible effect on the adsorption of pharmaceuticals to TMS-SBA-15 [17] Next, an amount of the adsorbent (10 mg) was added to the solution A sufficient volume of stock solution of pharmaceuticals was spiked to make a given initial concentration (100 g L−1 ) of each compound and to ensure the fraction of methanol in the solution was 5000 mL g−1 ), such as estrone, gemfibrozil, and ibuprofen (at pH 5.5), whereas less hydrophobic pharmaceuticals (acetaminophen, iopromide, and clofibric acid (see Table 1)) had low Kd values ( 0.05) as the pH increased from 5.5 to 7.6, whereas the Kd of anionic compounds (gemfibrozil, ibuprofen, diclofenac, ketoprofen, sulfamethoxazole, clofibric acid) decreased and the Kd of cationic compounds (trimethoprim, atenolol) increased in the same circumstance It is found that as pH increases from 5.5 to 7.6, the Kow of neutral compounds shows no change, whereas the Kow of anionic and cationic compounds get lower and higher, respectively (see Table 1) Not only did Kd and Kow for every pharmaceuticals vary in the same patterns as pH changed, but also the order of Kd for pharmaceuticals pH in the same group of compound followed the order of their Kow Fig The log–log relationship between pH-dependent octanol–water partition pH (Kow ) and adsorption coefficient (Kd ) for pharmaceuticals adsorption onto TMS-SBA15: (a) neutral and cationic compounds and (b) anionic compounds Solid lines show the linear regression fit (pH-dependent octanol-water coefficient) values at any pH (Fig 1): (i) estrone > carbamazepine > acetaminophen ≈ iopromide for neutral compounds; (ii) trimethoprim > atenolol for cationic compounds; (iii) gemfibrozil > ibuprofen > diclofenac > ketoprofen > sulfamethoxazole > clofibric acid for anionic compounds [17] A high level of agreement observed between the patterns pH of Kd and Kow suggests that hydrophobic interaction is a main interaction between TMS-SBA-15 and pharmaceuticals [17] However, it is more quantitative to verify whether there is a linear pH free energy relationship (LFER) occurring between Kd and Kow of pharmaceuticals [26] The linear log–log regressions (solid pH line) between Kd and Kow are presented in Fig Good linear correlations were observed for neutral and cationic compounds pH (log Kd = 0.45 log Kow + 2.82, r = 0.89) (Fig 2a) For anionic compounds, no good linear correlation could be generated for all compounds; instead, two linear regressions exist, corresponding to pH one group comprised solely of diclofenac (log Kd = 0.76 log Kow + 1.14, r = 1.00) and another group containing the other five compH pounds (log Kd = 0.97 log Kow + 1.91, r = 0.93) (Fig 2b) These observations support that the adsorption of pharmaceuticals to TMS-SBA-15 is primarily driven by hydrophobic interactions and is pH predictable by Kow values [26] The different regressions observed among the different group of compounds suggest that pharmaceutical adsorption involves implicit hydrophilic interactions [26] which might be varied significantly among the different groups T.X Bui et al / Journal of Hazardous Materials 254–255 (2013) 345–353 349 Table Adsorption coefficients (mL g−1 ) for individual pharmaceutical species toward TMSSBA-15 − Compounds KdHA KdA r2 FK0 b (%) Sulfamethoxazole Clofibric acid Ketoprofen Diclofenac Ibuprofen Gemfibrozil 386 ± 52a 16,707 ± 1197 24,574 ± 1620 82,186 ± 3924 206,416 ± 9833 505,895 ± 78,107 15 ± 26 0±4 49 ± 51 147 ± 101 469 ± 421 21,546 ± 6924 0.887 0.970 0.987 0.991 0.991 0.897 98 100 96 96 97 80 Compounds KdB KdHB r2 FK0 b (%) Atenolol Trimethoprim 11,261 ± 476 3227 ± 198 34 ± 33 ± 135 0.993 0.970 + 83c 75 a Standard error FK0 : Adsorption fraction of neutral species to the overall adsorption at pH 5.5, calculated by (KdHA ˛HA )/Kd for anionic compounds and (KdB ˛B )/Kd for cationic compounds.; c The value obtained at pH 7.6 b To confirm whether hydrophobic interaction is the main driving force responsible for the adsorption of pharmaceuticals onto TMS-SBA-15, desorption of pharmaceuticals from TMS-SBA-15 was determined using an organic solvent (herein ethanol was used) It was observed that nine compounds, except for clofibric acid, diclofenac, and ibuprofen showed complete desorption The desorption percentage of clofibric acid, diclofenac, and ibuprofen was 49.0, 91.4, and 76.2%, respectively These results support that hydrophobic interactions were major interactions between pharmaceuticals and TMS-SBA-15 [27] 3.2.2 Adsorption coefficient and pharmaceutical species existing in solution Unlike neutral pharmaceuticals that occur simply as the only form in aqueous solution, anionic and cationic pharmaceuticals exist as different forms in solution: anionic and neutral species for anionic compounds, cationic and neutral species for cationic compounds The individual species could have different adsorption coefficients to TMS-SBA-15 and have added disproportion contribution to the overall compound adsorption The adsorption edges were modeled by fitting with Eqs (3) and (4) to determine Kd values of individual species of anionic and cationic compounds (see Table 2) Although this model is empirical and does not indicate a mechanism of pharmaceutical interaction with TMS-SBA-15; fitting with model give insight into the relative contribution of pharmaceutical species to the overall compound adsorption Table suggests that the interaction between neutral species and the TMS-SBA-15 surface, as represented by KdHA for anionic compounds and KdB for cationic compounds, was more than 20 − times stronger than that of corresponding anions (KdA ) and cations + (KdHB ) As a result, neutral species accounted for a major contribution to the overall pharmaceutical adsorption even when the dominant species in solution phase was anions or cations At pH 5.5, the contribution of neutral species to the overall adsorption of five carboxylic pharmaceuticals ranged from 80% to 100% (Table 2), though the neutral species accounted for the fraction of 0.5% (clofibric acid) to 15.1% (gemfibrozil) [17] For cationic compounds, the fraction of neutral species was 1.5% for atenolol at pH 7.6 and was 2.8% for trimethoprim at pH 5.5 [17]; however, the species took responsibility for 83 and 75% of the overall adsorption of atenolol and trimethoprim, respectively, in such conditions The stronger interaction of neutral species to TMS-SBA-15 than for anions and cations provides another explanation for the decrease in the adsorption of anionic compounds and the increase in the adsorption of cationic compounds, as the pH increased Indeed, the increase of pH resulted in a lower number of neutral Fig Adsorption of pharmaceuticals onto TMS-SBA-15 as a function of contact time Error bars represent the standard deviation of sample replicates Reaction conditions: initial pharmaceutical concentration: 100 g L−1 , TMS-SBA-15: g L−1 , pH 5.5, at 25 ◦ C species but a higher amount of anions [17], leading a reduction in the overall adsorption of anionic pharmaceuticals In contrast, as the pH increased, cationic species turned into neutral species [17], resulting a higher overall adsorption of cationic compounds The adsorption of neutral compounds was believed to mainly rely on hydrophobic interaction because hydrogen bonding is not likely to be significant in aqueous phase [28], as supported by the fact that the adsorption of neutral hydrophilic compounds, such as acetaminophen and iopromide, onto TMS-SBA-15 was minimal (Fig 1) 3.3 Adsorption kinetics Adsorption kinetics is of great significance to evaluate the performance of a given adsorbent and gain insight into the underlying mechanisms Fitting measured data by adsorption kinetic models could determine how fast the adsorption occurs and generate parameters useful for calculation and modeling, while analyzing kinetic results with intraparticle diffusion equation could figure out rate-limiting steps, which are also helpful for controlling process in practical Therefore, experimental kinetic results of pharmaceutical adsorption onto TMS-SBA-15 have been modeled to obtain kinetic parameters and determine rate-limiting process 3.3.1 Adsorption kinetic parameters The adsorption results of pharmaceuticals to TMS-SBA-15 as a function of contact time are shown in Fig The adsorption of pharmaceuticals occurred dramatically in the initial 30 and then gradually reached equilibrium in h Adsorption kinetics of pharmaceuticals was examined using a nonlinear first-order model by Lagergren [29] (equation (5)) and a nonlinear pseudo-second-order model [30,31] (equation (6)): qt = qe (1 − e−k1 t ) qt = k2 q2e t + k2 qe t (5) (6) where k1 (h−1 ) and k2 (g g−1 h−1 ) are the equilibrium rate constant of first-order and pseudo-second-order models, respectively; qe is the amount of adsorbate adsorbed at equilibrium (g g−1 ); qt is the amount of adsorbate adsorbed at t (h) The nonlinear curve fitting of kinetic data was executed by employing nonlinear regression function of Sigma Plot version 11 (SPSS Inc., CA, USA) to obtain rate parameters A model gives a better fit if it results higher coefficients of determination (R2 ), lower standard errors of estimate (SE), and 350 T.X Bui et al / Journal of Hazardous Materials 254–255 (2013) 345–353 better agreement of calculated qe values with experimental qe values The kinetic parameters for pharmaceuticals obtained by fitting with first-order and pseudo-second-order equations are presented in Table The parameters obtained by fitting the adsorption data of acetaminophen, atenolol, clofibric acid, and iopromide are not shown because of extremely low coefficients of determination attained from the two models There were two types of fitting trends observed for the eight remaining pharmaceuticals Adsorption kinetic data of pharmaceuticals with high and moderate adsorption capacity, including gemfibrozil, ibuprofen, diclofenac, ketoprofen, estrone, carbamazepine, and trimethoprim showed a better fit with the pseudo-second-order model (Table 3) However, sulfamethoxazole, acetaminophen, atenolol, clofibric acid, and iopromide which had low adsorption capacity displayed a better compliance with the first-order model The different fitting trends among pharmaceuticals can be explained by the fact that pharmaceuticals with higher adsorption capacity usually adsorb more rapidly, as shown in detail below Thus, the faster pseudosecond-order model described experimental data more adequately for these pharmaceuticals Note that the adsorption of pharmaceuticals onto SBA-15 followed the pseudo-second-order equation as well [15] In the same group of properties (i.e anionic, neutral) a pharmaceuticals that had a higher adsorption capacity (qe ) exhibited a higher rate constant (k2 ), a faster initial reaction rate (h), and a shorter time required for the adsorption (t1/2 ) (see Table 3) The rate constant and adsorption capacity of pharmaceuticals followed the sequences: gemfibrozil > ibuprofen > diclofenac > ketoprofen > sulfamethoxazole for anionic compounds and estrone > carbamazepine for neutral compounds The inverse orders were observed for the adsorption time (t1/2 ) required for the pharmaceuticals The identical orders of the equilibrium adsorption amount (qe ) and the adsorption rate constant (k2 ) values for pharmaceuticals in the same group of properties in this study should be regarded specifically as a result from coadsorption of 12 pharmaceuticals The multiple-solute adsorption is driven strongly by competition among adsorbates, especially for adsorbates having similar properties In this study, 12 adsorbates have been divided into groups according to their physicochemical properties and adsorption behavior; then it is reasonable that one pharmaceuticals must mainly compete with the other compounds in the same group for preferred adsorption sites Having a higher equilibrium adsorption amount largely indicates that the pharmaceuticals is more competitive than other compounds in the same group A stronger competitor might adsorb more favorably and readily and then a higher amount of the compound could be accumulated onto the adsorbent surface during the same period; as such the more competitive adsorbate potentially has a higher adsorption rate constant In short, a pharmaceuticals exhibiting a higher equilibrium adsorption amount from a co-adsorption of multiple compounds probably has a higher adsorption rate constant than other compounds in the same group of properties 3.3.2 Rate-limiting steps It is well known that the adsorption process is composed of four consecutive steps [32]: (i) transport of adsorbate in the bulk solution; (ii) diffusion of adsorbate across the external boundary layer film of liquid surrounding the sorbent particles; (iii) diffusion of adsorbate either in the pores or along the pore walls; (iv) adsorption and desorption on the surface of the sorbent The rate-controlling step can be one or any combination of the steps For the adsorption occurring at a rapid mixing rate, the rate-limiting steps are very often limited to boundary layer (film) and intraparticle diffusion [33,34] To estimate the rate-limiting step, a quantitative approach of plotting qt versus t0.5 is generally employed [32–35], by which the intraparticle diffusion equation proposed by Weber and Morris [35] is examined: qt = ki t 0.5 (7) is the intraparticle diffusion rate parameter (g g−1 h−0.5 ) where ki If the plot is a straight line and passes through the origin, the adsorption process is only controlled by intraparticle diffusion; if not, more than one process influence the adsorption process The Weber–Morris plots of qt against t0.5 of pharmaceuticals, as shown in Fig 4, present multi-linearity [33] The results indicate that the adsorption was not only controlled by intraparticle diffusion but several steps were controlling the adsorption process [33,34] The first linear portion was attributed to the boundary layer diffusion [33,34] The second portion described the gradual adsorption stage, where intraparticle diffusion was rate controlled; this portion was ascribed to the diffusion of pharmaceuticals in mesopores [33] The third portion was the final equilibrium stage where the intraparticle diffusion started to slow down [34] and/or the diffusion of pharmaceutical molecules into micropores [33,34] which are present in the SBA-15 structure [36,37] Moreover, the third portion of the plots (Fig 4) was nearly parallel, suggesting that the rate of adsorption in the final equilibrium stage and/or diffusion in micropores was comparable for all pharmaceuticals In summary, the adsorption of pharmaceuticals onto TMS-SBA15 was initially controlled by the diffusion through the boundary Table Kinetic parameters obtained by fitting the adsorption kinetic data of pharmaceuticals onto TMS-SBA-15 with nonlinear first-order and pseudo-second-order kinetic equations Compounds First-order kinetic model −1 qe , cal (g g−1 ) R SE k2 (g g−1 h−1 ) qe , cal (g g−1 ) 45.166 19.786 6.648 3.931 0.441 98.23 86.49 77.98 60.35 21.75 0.681 0.553 0.680 0.671 0.969 0.590 5.420 9.985 9.747 1.655 4.694 0.461 0.111 0.0763 0.0201 98.49 89.13 83.81 66.12 24.64 16.150 4.082 91.36 80.11 0.777 0.835 4.605 8.602 0.349 0.0670 1.345 23.56 0.720 3.966 0.0813 qe , exp (g g−1 ) k1 (h Anionic compounds Gemfibrozil Ibuprofen Diclofenac Ketoprofen Sulfamethoxazole 98.32 90.75 87.40 70.92 20.38 Neutral compounds Estrone Carbamazepine 94.31 87.67 Cationic compounds Trimethoprim 26.65 −1 Pseudo-second-order kinetic model ) ◦ h (g g−1 h−1 ) R2 SE t1/2 (min) 45,538.4 3662.5 776.1 333.5 12.2 0.725 0.756 0.840 0.835 0.941 0.548 4.002 7.054 6.896 2.256 0.1 1.5 6.5 11.9 121.2 94.25 86.34 3102.2 499.4 0.852 0.935 3.753 5.409 1.8 10.4 25.15 51.4 0.808 3.286 29.3 −1 Conditions: adsorbent dose, g L ; pH, 5.5; temperature, 25 C; shaking rate, 200 rpm; initial pharmaceutical concentration, 100 g L ; qe , cal : calculated qe ; qe , exp : experimental qe ; SE: standard error of estimate.; h: initial reaction rate, calculated as h = k2 q2e ; t1/2 : the time taken for achieving 50% adsorption capacity; t1/2 (min) = k 1q × 60 e T.X Bui et al / Journal of Hazardous Materials 254–255 (2013) 345–353 351 Fig The Weber-Morris plots of intra-particle diffusion of (a) neutral and cationic pharmaceuticals and (b) anionic pharmaceuticals onto TMS-SBA-15 layer of liquid and subsequently limited by intraparticle diffusion including the diffusion of adsorbates in mesopores and micropores 3.4 Effect of adsorbent dose As the adsorbent dose was raised from 0.1 to g L−1 , the adsorption of pharmaceuticals onto TMS-SBA-15 generally increased (Fig 5) However, the increase in the adsorption efficiency of high adsorption-capacity pharmaceuticals was smaller than that of low adsorption-capacity compounds The increase of adsorbent dose from 0.1 to g L−1 has resulted a minimal increase of 1.6−8.4% in the adsorption percentages of gemfibrozil and estrone, whereas the same increase of adsorbent dose has inflated the adsorption percentages of carbamazepine, diclofenac, and ketoprofen 1.8–2.9 times As a result, the dose of 0.1 g L−1 of TMS-SBA-15 was essential to achieve the removal percentage of 96.7 and 86.9% for gemfibrozil and estrone, respectively, which were equivalent to 99.2 and 89.1% of the removal of the pharmaceuticals, respectively, at the dose of 1.0 g L−1 The dose of 0.2 g L−1 of TMS-SBA-15 was required to obtain 87.6 (ibuprofen), 74.0 (carbamazepine), 64.3 (diclofenac), and 59.9% (ketoprofen) of their removal at 1.0 g L−1 of the adsorbent dose Fig Effect of mass of adsorbent (TMS-SBA-15) on pharmaceutical adsorption Error bars represent the standard deviation of sample replicates Reaction conditions: initial pharmaceutical concentration: 100 g L−1 , pH 5.5, at 25 ◦ C, reaction time: 24 h 3.5 Effect of initial pharmaceutical concentration The influence of initial pharmaceutical concentration to the adsorption efficiency of TMS-SBA-15 was investigated by varying the concentration from 10 to 100 g L−1 The adsorbent dose of 0.1 g L−1 was used to ensure an adequate pharmaceutical concentration left in the solution after adsorption even for the samples spiked with a low initial concentration to avoid erroneous measurements The adsorption percentages of pharmaceuticals are shown in Fig The one-way analysis of variance (ANOVA) in conjunction with all pairwise multiple comparison procedures (Turkey Test) suggests that the initial pharmaceutical concentration did not result any statistically significant change (p > 0.05) in the adsorption percentages of all pharmaceuticals to TMS-SBA-15 The similar adsorption percentages were reported [12] for the adsorption of sulfamethoxazole onto carbonaceous adsorbents for two initial concentrations that differed by 2.5 orders of magnitude The results in this study suggest that adsorption results that are conducted at a high initial pharmaceutical concentration (100 g L−1 ) may represent for a range of different concentrations (10–100 g L−1 ) In addition, TMS-SBA15 showed comparable removal percentages of pharmaceuticals at different pharmaceutical concentrations, suggesting that the removal rate of a water treatment system used TMS-SBA-15 can be unchanged and resistant to the fluctuation of pharmaceutical concentration in the feeding influent Fig Effect of initial pharmaceutical concentration to the adsorption on TMSSBA-15 Error bars represent the standard deviation of sample replicates Reaction conditions: TMS-SBA-15: 0.1 g L−1 , pH 5.5, at 25 ◦ C, reaction time: 24 h 352 T.X Bui et al / Journal of Hazardous Materials 254–255 (2013) 345–353 boundary layer diffusion and intraparticle diffusion which includes mesopore diffusion and micropore diffusion Acknowledgement This research was supported by a grant of Environmental Fusion Technology Program from the Korea Environmental Protection Agency Appendix A Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jhazmat 2013.04.003 References Fig The adsorption percentages (%) of pharmaceuticals onto TMS-SBA-15 in the presence of Suwannee River NOM (SRNOM), Suwannee River FA (SRFA), and Suwannee River HA (SRHA), compared to the controls Error bars represent the standard deviation of sample replicates Reaction conditions: initial pharmaceutical concentration: 100 g L−1 , TMS-SBA-15: 0.2 g L−1 , pH 5.5, at 25 ◦ C, reaction time: 24 h 3.6 Effect of natural organic matter The adsorption percentages of pharmaceuticals in the presence of NOM were investigated and compared with control samples (Fig 7) The control experiments were conducted using the same working solution but without the addition of NOM The ANOVA test results indicate that NOM did not make any statistically significant change (p > 0.05) in the adsorption of all pharmaceuticals onto TMS-SBA-15 The results can be explained by the fact that NOM did not adsorb onto TMS-SBA-15 Indeed, our preliminary experiments showed that the adsorption of NOM onto TMS-SBA-15 was negligible The negligible adsorption of NOM could be due to electrostatic repulsion existing between NOM molecules and TMS-SBA-15 surfaces, which are all negatively charged [38,39]; nonetheless the TMS-SBA-15 surface has hydrophobic characteristics [40] Conclusions The present study aims to get insight into the adsorption of pharmaceuticals onto TMS-SBA-15 and to evaluate the influence of critical factors to the adsorption efficiency The results suggest that TMS-SBA-15 is an effective material for removal of pharmaceuticals in wastewater from drug manufacturers The adsorption onto TMS-SBA-15 was significantly affected by the solution pH and properties of pharmaceuticals including hydrophobicity and molecular charge (i.e neutral, anionic, and cationic) However, the adsorption efficiency of pharmaceuticals was not changed significantly under the changes of adsorbate concentration and was not influenced by the presence of NOM To remove extremely hydrophobic compounds, for example, gemfibrozil and estrone, a low adsorbent dose (0.1 g L−1 ) was needed in a short contact time (