Journal of Molecular Liquids 224 (2016) 151–157

Contents lists available at ScienceDirect

Journal of Molecular Liquids journal homepage: www.elsevier.com/locate/molliq

Adsorption of phosphorus from aqueous solution by cubic zeolitic imidazolate framework-8: Modeling, mechanical agitation versus sonication Mahmoud Shams a, Mohammad Hadi Dehghani a,b,⁎, Ramin Nabizadeh a,c, Alireza Mesdaghinia a, Mahmood Alimohammadi a, Ali Asghar Najafpoor d a

Tehran University of Medical Sciences, School of Public Health, Department of Environmental Health Engineering, Tehran, Islamic Republic of Iran Tehran University of Medical Sciences, Institute for Environmental Research, Center for Solid Waste Research, Tehran, Islamic Republic of Iran c Tehran University of Medical Sciences, Institute for Environmental Research, Center for Air Pollution Research, Tehran, Islamic Republic of Iran d Mashhad University of Medical Sciences, School of Health, Department of Environmental Health Engineering, Health Sciences Research Center, Mashhad, Islamic Republic of Iran b

a r t i c l e

i n f o

Article history: Received 15 August 2016 Received in revised form 18 September 2016 Accepted 19 September 2016 Available online 23 September 2016 Keywords: Metal-organic frameworks Zeolitic imidazolate framework-8 Phosphorus Adsorption Sonication

a b s t r a c t Cubic zeolitic imidazolate framework-8 (ZIF-8), a new class of hybrid adsorbent, was synthesized and investigated for phosphorus (P) removal from aqueous solution. A prediction model for P adsorption was developed by performing the experiments according to central composite design. The adsorption model showed that P adsorption is associated directly with time and ZIF-8 dosage and indirectly with initial P concentration. The removal also increased with decrease in pH until reaching the critical pH of about 2.6. The efﬁciency of P removal under mechanically stirred increased with agitation speed from 100 to 300 rpm. In contrast to high ultrasonic frequency (130 kHz), sonication under 35 kHz provides excellent dispersive mixing and shortened the equilibrium time. Study of coexisting ions showed that P adsorption hindered considerably in the presence of carbonate. Equilibrium studies indicated that P adsorbed onto ZIF-8 in monolayer and the adsorbent has deﬁnite localized sites that are energetically identical. The maximum Langmuir sorption capacity of ZIF-8 for P was 38.22 mg/g in the present study. Analysis of experimental data also revealed that the chemosorption is the rate limiting step in P adsorption by ZIF-8. In conclusion, the present report suggest zif-8 as an efﬁcient and fast removal sorbent for P from aqueous solutions. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Phosphorus (P) is well known as the key element in controlling the micro organisms growth and eutrophication of water bodies [1]. Agricultural and urban activities and a variety of industries such as glass and ceramic, chemical, and metal-plating industries, are responsible for discharge of P into natural water systems [2]. Since the late 1960s, when P removal from wastewater has received substantial attention, a large number of methods proposed by the scientists in attempt to enhance the P removal and minimize the operational issues in dephosphorization systems. Adsorption [3,4], precipitation [5,6], crystallization [7,8], ion exchange [9], and biological treatment [10] are among the successful technologies have been developed in this era. Adsorption, with a variety of advantages, including easy operation, low capital cost, low sludge production, effective in low concentrations of ⁎ Corresponding author at: Tehran University of Medical Sciences, School of Public Health, Department of Environmental Health Engineering, Tehran, Islamic Republic of Iran. E-mail address: [email protected] (M.H. Dehghani).

http://dx.doi.org/10.1016/j.molliq.2016.09.059 0167-7322/© 2016 Elsevier B.V. All rights reserved.

contaminant, recovering potential, and reliability of the process, received more attention in removing contaminants from water and wastewater streams [11]. Metal-organic frameworks (MOFS) are recognized in the recent years as highly promising materials for a variety of applications in catalyst, gas storage, clean energy, pharmacology, and sensing [12,13]. These extraordinary porous materials composed of metallic ions or metal clusters linked by organic ligands to form uniform three-dimensional networks [14,15]. MOFS attracted a great scientiﬁc interest due to their unique properties, including unusually high speciﬁc surface area, ultrahigh porosity, and tunable pore sizes and surface characteristics. Since the introduction of MOFS in 1990, the instability of most synthesized MOFs in the presence of moisture hindered their application in water environment purposes. The development of water-stable MOFS attracts attention for their utilization in the ﬁelds of water and wastewater. Zeolitic imidazilate frameworks (ZIFS), a subfamily of MOFS, consist of inorganic metal ions linked by organic imidazole ligands. ZIFs are known mainly because of the high thermal and chemical stability, very high surface area, and straightforward synthesis [16,17]. ZIF-8,

152

M. Shams et al. / Journal of Molecular Liquids 224 (2016) 151–157

the most studied compound in ZIFs family, constructed from tetrahedral zinc ions connected by imidazolate linkers to form a sodalite-type topology with pore size and window openings of 11.6 Å and 3.4 Å, respectively [18–20]. ZIF-8 was successfully used as adsorbent in the removal of arsenic [21,22], phthalic acid and diethyl phthalate [23], and p-arsanilic acid [24] from water solutions. However, to the best of our knowledge, there have been no reports in the literature regarding the P adsorption by ZIF-8 from aqueous solutions. During the last years, ultrasound has been effectively applied to improve a variety of processes, including synthesis, adsorption, extraction, advance oxidation process (AOP), etc. [25–28]. Sonication is one of the useful alternatives for mechanical mixing that accelerate the liquid turbulence through a variety of physical phenomena. It improves the mass transfer between liquid and solid phase by introducing the cycle of liquid compressing and expansion leading to the generation, grow and collapse of cavitation microbubbles [28– 30]. The design of the experiments was based on response surface methodology (RSM). RSM is a hybrid of mathematical and statistical technique that provides a statistical model for enhancement of the adsorption study by understanding the possible interactions among the variables and predict the optimal conditions for maximum adsorption, which are not determined in traditional one-factor design. Moreover, this methodology provide a useful tool in reducing the cost of chemicals and saving time by performing least number of experiments [31]. A multiple regression analysis of the experimental data is performed to evaluate the optimal mathematical model equation. The aim of this paper is to develop a prediction model for P adsorption onto ZIF-8 and study the inﬂuence of operating variables such as pH, coexisting ions and agitation type on P adsorption. A detailed report on adsorption kinetic and isotherm studies included in this paper. Batch experiments were conducted to investigate the inﬂuence of operating variables and simulation of adsorption kinetics

and isotherms. With the exception of the cases mentioned in the following, all the experiments were carried out at room temperature at 300 rpm and at pH 4. 2. Materials and methods 2.1. Preparation of ZIF-8 All chemicals used in synthesis of the adsorbent were of analytical grade and obtained from Sigma-Aldrich company. ZIF-8 was synthesized at ambient temperature according to the literature [32]. In brief, 0.594 g Zn(NO3)2·6H2O and 0.328 g 2-methylimidazole (Hmim) separately were added into 3 mL of deionized water and 3.76 g of ammonium hydroxide solution and well mixed to being completely dissolved. The zinc nitrate solution then added to Hmim solution, and the mixture was stirred for 10 min. The white color product separated using centrifuge in 3000 rpm for 10 min and washed several times with deionized water. To remove the water from ZIF-8 pores, the powder were ﬁnally dried overnight at 70 °C. The XRD pattern of the ZIF-8 is shown in Fig. 1(a). All of the diffraction patterns match well with those in the literature for ZIF-8, indicating that the sample has pure ZIF-8 phase. Fig. 1(b) shows the schematic of as-synthesized ZIF-8. SEM image of as-synthesized ZIF-8 using scanning electron microscope (SEM) is shown in Fig. 1(c). The images were clearly showed the uniform cubic shapes of the micrometer-sized crystals. 3. Results and discussion 3.1. Experimental design Central composite design (CCD), the most frequently used technique among RSM designs, was applied to design the experiment.

(a)

(b)

(c)

Fig. 1. Characteristics of the as-synthesized ZIF-8: (a) XRD pattern, (b) schematic of as-synthesized ZIF-8, and (c) SEM image of ZIF-8 crystals.

M. Shams et al. / Journal of Molecular Liquids 224 (2016) 151–157 Table 1 Experimental factors and coded levels of independent variables in this study.

Table 3 Estimated coefﬁcients of the ﬁtted polynomial model for P adsorption onto ZIF-8.

Variable level Factor

Variable

−1.68

−1

0

+1

+1.68

Time (min) ZIF-8 dose (g/L) P conc. (mg/L)

X1 X2 X3

2 100 5

13.8 282.4 7

31 550 10

48.2 817.6 13

60 1000 20

In this study, RSM was used to formulate the relationship between dependent (or response) variable deﬁned by Y and a set of independent variables deﬁned by X1, X2, X3,… Xn. The design matrix followed in the current study was made in random order by R software and was composed of ﬁve levels coded factors. The range and levels of individual variables were given in Table 1. As seen, time, ZIF − 8 dose and P concentration set at 2–60 min, 100– 1000 mg/L, and 5–20 mg/L, respectively. The coded values(xi) of the process parameters were determined by the following dimensionless equation: xi ¼

X i −X j ; δX

ð1Þ

where Xi is the actual value of independent variable, Xj is the actual value of independent variable at the center point, and δX is the step change. The experimental batch studies were performed according to CCD to describe the effects of different operating variables on P removal. As shown in Table 2, a total number of 20 experiments, involving 6 center points, 8 factorial points, and 6 axial points, was employed for the investigation. Based on responses obtained from the adsorption experiments, a quadratic equation model developed using the second degree polynomial equation expressed as k

k

i¼1

i¼1

153

k−1 k

Y ¼ β0 þ ∑ βi xi þ ∑ βii x2i þ ∑ ∑ βij xi x j þ ε

ð2Þ

i¼1 j−1

where Y, β0, βi, βii, and βij are the predicted dependent variable (P adsorption), the intercept, the linear effect coefﬁcient, the quadratic effect coefﬁcient, and the interaction effect coefﬁcient between factor i and j that estimated by the model. ε is the residual term, and xi and xj are independent parameters [33].

Coefﬁcient estimate

SE

t-value

p-value

Intercept X1 X2 X3 X1 X2 X 1 X3 X 2 X3 X12 X22 X23

77.08 4.69 4.01 −4.97 −1.6 −0.55 1.12 −1.52 −2.23 0.48

0.862 0.569 0.569 0.569 0.746 0.746 0.746 0.547 0.547 0.547

89.41 8.24 7.05 −8.73 −2.14 −0.73 1.50 −2.78 −4.08 0.87

7.4 × 10−16 9 × 10−6 3.4 × 10−5 5.3 × 10−6 0.057 0.478 0.162 0.019 0.002 0.400

involving 3 main effects, 3 quadratic effects, and 3 interaction effects as follows:

Y ¼ 77:08 þ 4:69X 1 þ 4:01X 2 −4:97X 3 −1:52X 21 −2:23X 22 þ 0:48X 23 −1:6X 1 X 2 −0:55X 1 X 3 þ 1:12X 2 X 3

ð3Þ

Eq. (3) suggests that the P removal efﬁciency increased with the time (X1) and ZIF-8 dose (X2) and decreased P concentration (X3). ANOVA was used to evaluate the signiﬁcance of the developed model. For a valid statistical model, adjusted R2 should be within R2 ± 0.2. As shown in Table 4, the difference between R2 and Radj is about 0.04. In addition, the Lack of Fit value is 0.22, which is not significant relative to the pure error. These values indicate that the adsorption behavior could be explained well by the predicted model. Fig. 2 plot the P adsorption efﬁciencies of ZIF-8 observed in experimental studies versus those predicted by the quadratic model (Eq. (3)) and presented in Table 2. As seen in Fig. 2, the ﬁtted values of the model is close to real values obtained in the experiments. The good agreement between these values conﬁrm the preciseness of developed model for prediction of P adsorption by ZIF-8.

Table 4 Analysis of variance (ANOVA) for the ﬁtted polynomial model for P adsorption onto ZIF-8.

3.2. Model ﬁtting and statistical analysis Table 3 summarized the analysis of variance (ANOVA) of the experimental data presented in Table 2. By applying multiple regression analysis on coded data, a quadratic regression equation was developed

Model term

Model formula

df

Sum Sq

Mean Sq

F value

Pr (NF)

First-order response Second-order response Pure quadratic response Residuals Lack of ﬁt Pure error

3 3 3 10 5 5

866.09 33.03 110.19 44.63 29.99 14.64

288.69 11.00 36.72 4.46 5.99 2.92

64.68 2.46 8.22

7.4 × 10–7 0.12 4 × 10–3

2.04

0.22

Notes: multiple R2: 0.957; adjusted R2: 0.919; F-statistic: 25.13 on 9 and 10 DF; p-value: 0.0000.

Table 2 The design of experiments and experimental responses for P adsorption by ZIF-8. Run order

Standard order

1 2 3 4 5 6 7 8 9 10

2 4 6 14 12 11 3 15 18 19

Coded variable

Response (removal)

X1

X2

X3

observed

predicted

1 1 1 0 0 0 −1 0 0 0

−1 1 −1 0 1.7 −1.7 1 0 0 0

−1 −1 1 1.7 0 0 −1 0 0 0

82.9 84.0 71.5 67.3 78.7 61.8 76.5 75.4 79.3 75.1

82.7 85.3 69.5 70.1 77.5 64.0 78.0 77.1 77.1 77.1

Run order

Standard order

11 12 13 14 15 16 17 18 19 20

13 10 16 7 8 9 5 20 17 1

Coded variable

Response (removal)

X1

X2

X3

observed

predicted

0 1.7 0 −1 1 −1.7 −1 0 0 −1

0 0 0 1 1 0 −1 0 0 −1

−1.7 0 0 1 1 0 1 0 0 −1

88.9 79.9 77.0 71.8 77.3 64.7 59.8 78.8 77.0 68.8

86.8 80.7 77.1 71.4 76.5 64.9 58.0 77.1 77.1 69.1

154

M. Shams et al. / Journal of Molecular Liquids 224 (2016) 151–157

Fig. 2. Experimental P adsorption versus predicted removal efﬁciencies. Fig. 4. The effect of pH on P adsorption. P: 10 mg/L; contact time: 30 min; ZIF-8: 0.5 g/L.

Fig. 3. Contour and 3D response surface plots for P removal. The effect of (a, b) initial P concentration and time, (c, d) ZIF-8 dose and time.

M. Shams et al. / Journal of Molecular Liquids 224 (2016) 151–157

155

100 Removal (%)

35 kHz 80

130 kHz

60

300 rpm 200 rpm

40

100 rpm

20 0 0

10

20

30 40 Time (min)

50

60

70

Fig. 7. Comparison of agitation by mechanical mixer and ultrasound at different conditions on P adsorption. P: 10 mg/L; ZIF-8: 0.5 g/L.

Fig. 5. The effect of coexisting ions on P adsorption. P: 10 mg/L, contact time: 30 min. ZIF-8: 0.5 g/L.

3.3. Effect of the operating variables on P adsorption Graphical representation of the model developed in the present work (Eq. (3)) is shown in Fig. 3(a–d). Contour plots and 3D response surface plots were used to visualize the independent variables effect to the response. These plots provide facile method in ﬁnding the optimum points at which the desired response occurred. P removal as a function of initial P concentration and time shown in Fig. 3(a, b). As expected, P adsorption decreased with increasing initial P concentration. This is expected based on previously reports [21,22]. The higher P concentrations in the solution hinders the P ions to adsorb onto ZIF-8. Fig. 3(c, d) shows the P adsorption increased by ZIF-8 to reach an optimum value. The initial increasing is a consequence of higher available adsorption sites at higher adsorbent doses. Increasing the adsorbent dose higher than optimum value decreased removal efﬁciency. This could attributed to the increasing pH after ZIF-8 added to solution, especially when higher doses were used.

3.4. Effect of pH The effect of pH on P adsorption in the range of 2–12 was evaluated separately in a set of batch experiments. As can be seen in Fig. 4, the P removal increased with decrease in initial solution pH and the maximum adsorption occurred at pH 2.8. This can be explained on the basis of ZIF-8 surface charge. Zero point of charge (pHzpc) of ZIF-8 reported around 9.5 in the literature [22].

(a)

(b)

Fig. 6. ZIF-8 dispersion by agitation under (a) mechanical mixer and (b) ultrasound after 15 min.

The charge of the adsorbent surface getting more positive with decreasing the solution pH and P adsorption increased as a result of higher electrostatic interaction. The decrease in P removal efﬁciency could also attributed to the adsorption competition between P− and OH– ions especially in higher pH. The decrease in percent adsorption by increasing the pH was also reported in the literature on adsorption of arsenic and phthalic acid [23]. Although P adsorption increases under acidic condition, decreasing the solution pH below 2.6 cause a sharp reduction in P adsorption. The sharp fall in P removal could attributed to the instability of ZIF-8 under strong acidic conditions. When pH of the solution adjusted below around 2.6, the solution turbidity disappears gradually due to the dissolution of ZIF-8.

3.5. Effect of competing ions The adsorption of P on ZIF-8 in presence of coexisting anions was studied using different initial concentrations of sulfate, chloride, nitrate, carbonate, and bicarbonate. As shown in Fig. 5, the inﬂuence of the applied anions other than carbonate is negligible for the selected concentrations. The high interference of carbonate on P adsorption could be attributed to changes occurred pH of the solution. In the presence of 200 and 400 mg/L carbonate, the pH level of the solution increased from 5.7 to 8.5 and 9, respectively. The surface charge of ZIF-8 become more negative at higher pH and this is led to a static repulsion of PO34 and HPO24 −. Previous reports also indicated that the carbonate ions can adsorbed by zinc and interfere with the adsorption of other negatively charged ions [21].

3.6. Effect of agitation speed and sonication To study the effect of mixing speed and sonication on P adsorption, the kinetic experiments were carried out under mechanical stirring at 100, 200, and 300 rpm and in ultrasonic bath at 35 and 130 kHz at constant temperature. Fig. 6(a, b) compared samples agitated by conventional mixer and sonication for the same time. As can be seen, the shear forces produced during the microbubble collapsing provide sufﬁcient energy for breaking adsorbent agglomerates and generating a well dispersive mixing system. A comparison of agitation by mechanical mixer and ultrasound at different conditions on P removal efﬁciency is shown in Fig. 7. It can be deduced from Fig. 7 that the increasing turbulence at higher mixing degrees decreases the boundary layer thickness of solid–liquid phase and causes a higher mass transfer rate. According to the Fig. 7, it can also deduced that the equilibrium time can decreased signiﬁcantly when the solution irradiated with a frequency of 35 kHz. Sonication with operating frequency of 130 kHz, on the other hand, was not effective for mixing purpose in the studied time.

156

M. Shams et al. / Journal of Molecular Liquids 224 (2016) 151–157

Table 5 Isotherm models and the parameters of ﬁtted models. Isotherm

Formula

Linear form 1 1 ¼ qm Ce þ qmb

Plot

Parameter

Ce qe

qmax (mg/g) KL (L/mg) R2 KF (mg/g(L/mg)1/n) n R2 kt (L/mg) B1 R2 qmax (mg/g) β R2

Langmuir

qe ¼

Freundlich

qe = KFC1/n e

Log qe = log KF + 1n log Ce

logqe vs. logCe

Temkin

; ln ðkT C e Þ qe ¼ RT b

qe = B1 ln .kt + B1 ln Ce

qe vs. lnCe

Dubinin–Radushkevich

qe =qmexp. (−βε2)

lnqe = lnqm − βϵ2

qe vs. ε2

Ce/qe

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0

(a)

0

Ln qe

Ce qe

qm bC e 1þbC e

10

30

Ce

40

(b)

Adsorption isotherms are important part of a sorption study that describe the pollutants interaction with the adsorbents. Equilibrium relationships also provide information about the adsorbents capacities and their surface properties, which is necessary for designing real sorption systems [34]. Langmuir, Freundlich, Dubinin–Radushkevich, and Temkin models were used to simulate the equilibrium data obtained from P adsorption onto ZIF-8 after 12 h agitation. Table 5 listed the ﬁtted isotherm models, their linear forms, and the values of parameters obtained by ﬁtting the experimental data with the models and shown in Fig. 8(a–d). As it seen, the value of R2 is higher for Langmuir model than the other models. The adsorption theory of Langmuir model is based on the kinetic principles and it proposes the monolayer surface adsorption on the ideal solid with deﬁnite localized sites that are energetically identical. As seen in Table 5, the maximum monolayer adsorption capacity of ZIF-8 for P was 38.22 mg/g in the present work. 3.8. Kinetic modeling

-2

0

4

Ln Ce2

4

(c)

3.5 3 Ln qe

38.22 0.94 0.99 640.3 3.57 0.93 0.59 5.86 0.97 31.1 5.07 0.87

3.7. Equilibrium study

20

4 3.5 3 2.5 2 1.5 1 0.5 0

vs. Ce

2.5 2 1.5

Kinetic studies provide required information on selecting optimum operating conditions in real batch sorption systems. In the present work, the experimental data were ﬁtted to three of the most widely used kinetic models as listed in Table 6 and shown in Fig. 9(a–c). Table 7 shows the ﬁtted experimental data with the pseudo-ﬁrstorder, pseudo-second-order, and intraparticle diffusion kinetic models. As seen, pseudo-second-order kinetic model with higher R2 values showed better compliance with the experimental data, suggesting the adsorption to be controlled by chemosorption.

1 4. Conclusion

0.5 0 0

10

20 2

50 (d)

qe

40 30 20 10 0 -2

-1

0

1 Ln Ce

2

3

4

Fig. 8. Fitting the experimental data with (a) Langmuir, (b) Freundlich, (c) Dubinin– Radushkevich, and (d) Temkin models.

The cubic ZIF-8 synthesized and applied for P removal from aqueous solutions. According to central composite design, a quadratic model was developed for prediction of P removal by ZIF-8. The prediction model suggest that the P removal increased by ZIF-8 dose and decreased with increasing initial P concentration. P removal also increased with decrease in initial solution pH and the maximum adsorption occurred around pH 2.8. Below the optimum pH, P concentration increased sharply due to the dissolution of ZIF-8 in strong acidic condition. Carbonate ions, due to increasing pH, showed a high level of interference on P adsorption. A higher mass transfer rate and removal efﬁciency was obtained when the solution agitated at higher mechanical speed and when the solution agitated by sonication with a frequency of 35 kHz. Present study revealed that the adsorption were ﬁtted well by Langmuir model with a maximum monolayer adsorption capacity of 38.22 mg/g. Pseudo-second-order kinetic model described the experimental data well, suggesting the adsorption to be controlled by chemosorption. Present report suggest ZIF-8 as an efﬁcient and fast removal sorbent for P from aqueous solutions.

M. Shams et al. / Journal of Molecular Liquids 224 (2016) 151–157

157

Table 6 Kinetic models ﬁtted with the experimental data. Kinetic model

Formula

Plot

Pseudo-ﬁrst-order kinetic model

Logðqe −qt Þ ¼

Pseudo-second-order kinetic model

t qt

Intraparticle diffusion kinetic model

qt = kp . t0.5 + c

log(qe − qt) vs. t

k₁ ; logqe − 2:303 :t

t qt

1 1 ¼ k₂qе 2 þ qе :t

vs. t

qt vs. t0.5

Acknowledgment

(a)

The authors are grateful to express their appreciate for the help and support of the laboratory staff members of environmental health department of Tehran University of Medical Sciences.

References

(b)

(c)

Fig. 9. Fitting the experimental data with (a) pseudo-ﬁrst-order, (b) pseudo-second-order, and (c) intraparticle diffusion kinetic models.

[1] Z. Zhu, H. Zeng, Y. Zhu, F. Yang, H. Zhu, H. Qin, W. Wei, Sep. Purif. Technol. 117 (2013) 124–130. [2] M. Shams, S. Dobaradaran, B. Ramavandi, M. Qasemi, M. Afsharnia, Fresenius Environ. Bull. 22 (2013) 722–726. [3] J. Ye, X. Cong, P. Zhang, G. Zeng, E. Hoffmann, Y. Wu, H. Zhang, W. Fang, J. Mol. Liq. 216 (2016) 35–41. [4] W. Gu, Q. Xie, C. Qi, L. Zhao, D. Wu, Powder Technol. 301 (2016) 723–729. [5] C. Han, Z. Wang, W. Yang, Q. Wu, H. Yang, X. Xue, Ecol. Eng. 89 (2016) 1–6. [6] S. Yang, P. Jin, X. Wang, Q. Zhang, X. Chen, Chem. Eng. J. 292 (2016) 246–254. [7] S. Kaneko, K. Nakajima, J. Water Pollut. Control Fed. 60 (1988) 1239–1244. [8] D. Bian, S. Ai, J. Liu, Y. Zuo, X. Tian, J. Environ. Sci. 23 (2011) S106–S109. [9] Y.I. Seo, K.H. Hong, S.H. Kim, D. Chang, K.H. Lee, Y.D. Kim, J. Ind. Eng. Chem. 19 (2013) 744–747. [10] R.J. Seviour, T. Mino, M. Onuki, FEMS Microbiol. Rev. 27 (2003) 99–127. [11] M. Shams, M. Qasemi, S. Dobaradaran, A.H. Mahvi, Fresenius Environ. Bull. 22 (2013) 2604–2609. [12] M.E. Hailian Li, M. O'Keeffe, O.M. Yaghi, Nature 402 (1999) 276–279. [13] H.C. Zhou, J.R. Long, O.M. Yaghi, Chem. Rev. 112 (2012) 673–674. [14] F. Tan, M. Liu, K. Li, Y. Wang, J. Wang, X. Guo, G. Zhang, C. Song, Chem. Eng. J. 281 (2015) 360–367. [15] M. Massoudinejad, M. Ghaderpoori, A. Shahsavani, M.M. Amini, J. Mol. Liq. 221 (2016) 279–286. [16] Y. Pan, W. Liu, Y. Zhao, C. Wang, Z. Lai, J. Membr. Sci. 493 (2015) 88–96. [17] S. Wang, Y. Fan, X. Jia, Chem. Eng. J. 256 (2014) 14–22. [18] F. Cacho-Bailo, B. Seoane, C. Téllez, J. Coronas, J. Membr. Sci. 464 (2014) 119–126. [19] E.L. Bustamante, J. Fernández, J.M. Zamaro, J. Colloid Interface Sci. 424 (2014) 37–43. [20] J. Duan, Y. Pan, F. Pacheco, E. Litwiller, Z. Lai, I. Pinnau, J. Membr. Sci. 476 (2015) 303–310. [21] M. Jian, B. Liu, G. Zhang, R. Liu, X. Zhang, J. Colloid Interface Sci. 465 (2015) 67–76. [22] B. Liu, M. Jian, R. Liu, J. Yao, X. Zhang, J. Colloid Interface Sci. 481 (2015) 358–366. [23] N.A. Khan, B.K. Jung, Z. Hasan, S.H. Jhung, J. Hazard. Mater. 282 (2015) 194–200. [24] B.K. Jung, J.W. Jun, Z. Hasan, S.H. Jhung, Chem. Eng. J. 267 (2015) 9–15. [25] S. Zinatloo-Ajabshir, M. Salavati-Niasari, J. Mol. Liq. 216 (2016) 545–551. [26] G. Ameta, A.K. Pathak, C. Ameta, R. Ameta, P.B. Punjabi, J. Mol. Liq. 211 (2015) 934–937. [27] S. Agarwal, I. Tyagi, V.K. Gupta, M.H. Dehghani, A. Bagheri, K. Yetilmezsoy, A. Amrane, B. Heibati, S. Rodriguez-Couto, J. Mol. Liq 221 (2016) 1237–1242. [28] S.R. Pouran, A. Bayrami, A.R.A. Aziz, W.M.A.W. Daud, M.S. Shafeeyan, J. Mol. Liq 222 (2016) 1076–1084. [29] H. Mazaheri, M. Ghaedi, A. Asfaram, S. Hajati, J. Mol. Liq. 219 (2016) 667–676. [30] M. Rajabi, S. Asemipour, B. Barﬁ, M.R. Jamali, M. Behzad, J. Mol. Liq. 194 (2014) 166–171. [31] R. Jayakumar, M. Rajasimman, C. Karthikeyan, Ecotoxicol. Environ. Saf. 121 (2015) 199–210. [32] M. He, J. Yao, Q. Liu, K. Wang, F. Chen, H. Wang, Microporous Mesoporous Mater. 184 (2014) 55–60. [33] P.F. de Sales, Z.M. Magriotis, M.A.L.S. Rossi, R.F. Resende, C.A. Nunes, J. Environ. Manag. 151 (2015) 144–152. [34] K.Y. Foo, B.H. Hameed, Chem. Eng. J. 156 (2010) 2–10.

Table 7 Constants obtained from kinetic models for P adsorption onto ZIF-8. Pseudo-ﬁrst order

Pseudo-second order

C0 [mg/L]

qe, exp. [mg/g]

qe,cal [mg/g]

K1 [min

5 10 15 20

9.52 18.64 29.36 35.76

3.8 10.59 20.91 27.61

0.06 0.08 0.07 0.054

−1

]

2

R

qe,cal [mg/g]

K2 [min

0.96 0.98 0.97 0.98

9.78 19.45 32.37 37.65

0.04 0.02 0.005 0.003

Intraparticle diffusion −1

]

R

Kp [mg/g min−0.5]

R2

0.99 0.99 0.99 0.99

0.52 1.42 3.11 3.83

0.80 0.86 0.89 0.96

2

Contents lists available at ScienceDirect

Journal of Molecular Liquids journal homepage: www.elsevier.com/locate/molliq

Adsorption of phosphorus from aqueous solution by cubic zeolitic imidazolate framework-8: Modeling, mechanical agitation versus sonication Mahmoud Shams a, Mohammad Hadi Dehghani a,b,⁎, Ramin Nabizadeh a,c, Alireza Mesdaghinia a, Mahmood Alimohammadi a, Ali Asghar Najafpoor d a

Tehran University of Medical Sciences, School of Public Health, Department of Environmental Health Engineering, Tehran, Islamic Republic of Iran Tehran University of Medical Sciences, Institute for Environmental Research, Center for Solid Waste Research, Tehran, Islamic Republic of Iran c Tehran University of Medical Sciences, Institute for Environmental Research, Center for Air Pollution Research, Tehran, Islamic Republic of Iran d Mashhad University of Medical Sciences, School of Health, Department of Environmental Health Engineering, Health Sciences Research Center, Mashhad, Islamic Republic of Iran b

a r t i c l e

i n f o

Article history: Received 15 August 2016 Received in revised form 18 September 2016 Accepted 19 September 2016 Available online 23 September 2016 Keywords: Metal-organic frameworks Zeolitic imidazolate framework-8 Phosphorus Adsorption Sonication

a b s t r a c t Cubic zeolitic imidazolate framework-8 (ZIF-8), a new class of hybrid adsorbent, was synthesized and investigated for phosphorus (P) removal from aqueous solution. A prediction model for P adsorption was developed by performing the experiments according to central composite design. The adsorption model showed that P adsorption is associated directly with time and ZIF-8 dosage and indirectly with initial P concentration. The removal also increased with decrease in pH until reaching the critical pH of about 2.6. The efﬁciency of P removal under mechanically stirred increased with agitation speed from 100 to 300 rpm. In contrast to high ultrasonic frequency (130 kHz), sonication under 35 kHz provides excellent dispersive mixing and shortened the equilibrium time. Study of coexisting ions showed that P adsorption hindered considerably in the presence of carbonate. Equilibrium studies indicated that P adsorbed onto ZIF-8 in monolayer and the adsorbent has deﬁnite localized sites that are energetically identical. The maximum Langmuir sorption capacity of ZIF-8 for P was 38.22 mg/g in the present study. Analysis of experimental data also revealed that the chemosorption is the rate limiting step in P adsorption by ZIF-8. In conclusion, the present report suggest zif-8 as an efﬁcient and fast removal sorbent for P from aqueous solutions. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Phosphorus (P) is well known as the key element in controlling the micro organisms growth and eutrophication of water bodies [1]. Agricultural and urban activities and a variety of industries such as glass and ceramic, chemical, and metal-plating industries, are responsible for discharge of P into natural water systems [2]. Since the late 1960s, when P removal from wastewater has received substantial attention, a large number of methods proposed by the scientists in attempt to enhance the P removal and minimize the operational issues in dephosphorization systems. Adsorption [3,4], precipitation [5,6], crystallization [7,8], ion exchange [9], and biological treatment [10] are among the successful technologies have been developed in this era. Adsorption, with a variety of advantages, including easy operation, low capital cost, low sludge production, effective in low concentrations of ⁎ Corresponding author at: Tehran University of Medical Sciences, School of Public Health, Department of Environmental Health Engineering, Tehran, Islamic Republic of Iran. E-mail address: [email protected] (M.H. Dehghani).

http://dx.doi.org/10.1016/j.molliq.2016.09.059 0167-7322/© 2016 Elsevier B.V. All rights reserved.

contaminant, recovering potential, and reliability of the process, received more attention in removing contaminants from water and wastewater streams [11]. Metal-organic frameworks (MOFS) are recognized in the recent years as highly promising materials for a variety of applications in catalyst, gas storage, clean energy, pharmacology, and sensing [12,13]. These extraordinary porous materials composed of metallic ions or metal clusters linked by organic ligands to form uniform three-dimensional networks [14,15]. MOFS attracted a great scientiﬁc interest due to their unique properties, including unusually high speciﬁc surface area, ultrahigh porosity, and tunable pore sizes and surface characteristics. Since the introduction of MOFS in 1990, the instability of most synthesized MOFs in the presence of moisture hindered their application in water environment purposes. The development of water-stable MOFS attracts attention for their utilization in the ﬁelds of water and wastewater. Zeolitic imidazilate frameworks (ZIFS), a subfamily of MOFS, consist of inorganic metal ions linked by organic imidazole ligands. ZIFs are known mainly because of the high thermal and chemical stability, very high surface area, and straightforward synthesis [16,17]. ZIF-8,

152

M. Shams et al. / Journal of Molecular Liquids 224 (2016) 151–157

the most studied compound in ZIFs family, constructed from tetrahedral zinc ions connected by imidazolate linkers to form a sodalite-type topology with pore size and window openings of 11.6 Å and 3.4 Å, respectively [18–20]. ZIF-8 was successfully used as adsorbent in the removal of arsenic [21,22], phthalic acid and diethyl phthalate [23], and p-arsanilic acid [24] from water solutions. However, to the best of our knowledge, there have been no reports in the literature regarding the P adsorption by ZIF-8 from aqueous solutions. During the last years, ultrasound has been effectively applied to improve a variety of processes, including synthesis, adsorption, extraction, advance oxidation process (AOP), etc. [25–28]. Sonication is one of the useful alternatives for mechanical mixing that accelerate the liquid turbulence through a variety of physical phenomena. It improves the mass transfer between liquid and solid phase by introducing the cycle of liquid compressing and expansion leading to the generation, grow and collapse of cavitation microbubbles [28– 30]. The design of the experiments was based on response surface methodology (RSM). RSM is a hybrid of mathematical and statistical technique that provides a statistical model for enhancement of the adsorption study by understanding the possible interactions among the variables and predict the optimal conditions for maximum adsorption, which are not determined in traditional one-factor design. Moreover, this methodology provide a useful tool in reducing the cost of chemicals and saving time by performing least number of experiments [31]. A multiple regression analysis of the experimental data is performed to evaluate the optimal mathematical model equation. The aim of this paper is to develop a prediction model for P adsorption onto ZIF-8 and study the inﬂuence of operating variables such as pH, coexisting ions and agitation type on P adsorption. A detailed report on adsorption kinetic and isotherm studies included in this paper. Batch experiments were conducted to investigate the inﬂuence of operating variables and simulation of adsorption kinetics

and isotherms. With the exception of the cases mentioned in the following, all the experiments were carried out at room temperature at 300 rpm and at pH 4. 2. Materials and methods 2.1. Preparation of ZIF-8 All chemicals used in synthesis of the adsorbent were of analytical grade and obtained from Sigma-Aldrich company. ZIF-8 was synthesized at ambient temperature according to the literature [32]. In brief, 0.594 g Zn(NO3)2·6H2O and 0.328 g 2-methylimidazole (Hmim) separately were added into 3 mL of deionized water and 3.76 g of ammonium hydroxide solution and well mixed to being completely dissolved. The zinc nitrate solution then added to Hmim solution, and the mixture was stirred for 10 min. The white color product separated using centrifuge in 3000 rpm for 10 min and washed several times with deionized water. To remove the water from ZIF-8 pores, the powder were ﬁnally dried overnight at 70 °C. The XRD pattern of the ZIF-8 is shown in Fig. 1(a). All of the diffraction patterns match well with those in the literature for ZIF-8, indicating that the sample has pure ZIF-8 phase. Fig. 1(b) shows the schematic of as-synthesized ZIF-8. SEM image of as-synthesized ZIF-8 using scanning electron microscope (SEM) is shown in Fig. 1(c). The images were clearly showed the uniform cubic shapes of the micrometer-sized crystals. 3. Results and discussion 3.1. Experimental design Central composite design (CCD), the most frequently used technique among RSM designs, was applied to design the experiment.

(a)

(b)

(c)

Fig. 1. Characteristics of the as-synthesized ZIF-8: (a) XRD pattern, (b) schematic of as-synthesized ZIF-8, and (c) SEM image of ZIF-8 crystals.

M. Shams et al. / Journal of Molecular Liquids 224 (2016) 151–157 Table 1 Experimental factors and coded levels of independent variables in this study.

Table 3 Estimated coefﬁcients of the ﬁtted polynomial model for P adsorption onto ZIF-8.

Variable level Factor

Variable

−1.68

−1

0

+1

+1.68

Time (min) ZIF-8 dose (g/L) P conc. (mg/L)

X1 X2 X3

2 100 5

13.8 282.4 7

31 550 10

48.2 817.6 13

60 1000 20

In this study, RSM was used to formulate the relationship between dependent (or response) variable deﬁned by Y and a set of independent variables deﬁned by X1, X2, X3,… Xn. The design matrix followed in the current study was made in random order by R software and was composed of ﬁve levels coded factors. The range and levels of individual variables were given in Table 1. As seen, time, ZIF − 8 dose and P concentration set at 2–60 min, 100– 1000 mg/L, and 5–20 mg/L, respectively. The coded values(xi) of the process parameters were determined by the following dimensionless equation: xi ¼

X i −X j ; δX

ð1Þ

where Xi is the actual value of independent variable, Xj is the actual value of independent variable at the center point, and δX is the step change. The experimental batch studies were performed according to CCD to describe the effects of different operating variables on P removal. As shown in Table 2, a total number of 20 experiments, involving 6 center points, 8 factorial points, and 6 axial points, was employed for the investigation. Based on responses obtained from the adsorption experiments, a quadratic equation model developed using the second degree polynomial equation expressed as k

k

i¼1

i¼1

153

k−1 k

Y ¼ β0 þ ∑ βi xi þ ∑ βii x2i þ ∑ ∑ βij xi x j þ ε

ð2Þ

i¼1 j−1

where Y, β0, βi, βii, and βij are the predicted dependent variable (P adsorption), the intercept, the linear effect coefﬁcient, the quadratic effect coefﬁcient, and the interaction effect coefﬁcient between factor i and j that estimated by the model. ε is the residual term, and xi and xj are independent parameters [33].

Coefﬁcient estimate

SE

t-value

p-value

Intercept X1 X2 X3 X1 X2 X 1 X3 X 2 X3 X12 X22 X23

77.08 4.69 4.01 −4.97 −1.6 −0.55 1.12 −1.52 −2.23 0.48

0.862 0.569 0.569 0.569 0.746 0.746 0.746 0.547 0.547 0.547

89.41 8.24 7.05 −8.73 −2.14 −0.73 1.50 −2.78 −4.08 0.87

7.4 × 10−16 9 × 10−6 3.4 × 10−5 5.3 × 10−6 0.057 0.478 0.162 0.019 0.002 0.400

involving 3 main effects, 3 quadratic effects, and 3 interaction effects as follows:

Y ¼ 77:08 þ 4:69X 1 þ 4:01X 2 −4:97X 3 −1:52X 21 −2:23X 22 þ 0:48X 23 −1:6X 1 X 2 −0:55X 1 X 3 þ 1:12X 2 X 3

ð3Þ

Eq. (3) suggests that the P removal efﬁciency increased with the time (X1) and ZIF-8 dose (X2) and decreased P concentration (X3). ANOVA was used to evaluate the signiﬁcance of the developed model. For a valid statistical model, adjusted R2 should be within R2 ± 0.2. As shown in Table 4, the difference between R2 and Radj is about 0.04. In addition, the Lack of Fit value is 0.22, which is not significant relative to the pure error. These values indicate that the adsorption behavior could be explained well by the predicted model. Fig. 2 plot the P adsorption efﬁciencies of ZIF-8 observed in experimental studies versus those predicted by the quadratic model (Eq. (3)) and presented in Table 2. As seen in Fig. 2, the ﬁtted values of the model is close to real values obtained in the experiments. The good agreement between these values conﬁrm the preciseness of developed model for prediction of P adsorption by ZIF-8.

Table 4 Analysis of variance (ANOVA) for the ﬁtted polynomial model for P adsorption onto ZIF-8.

3.2. Model ﬁtting and statistical analysis Table 3 summarized the analysis of variance (ANOVA) of the experimental data presented in Table 2. By applying multiple regression analysis on coded data, a quadratic regression equation was developed

Model term

Model formula

df

Sum Sq

Mean Sq

F value

Pr (NF)

First-order response Second-order response Pure quadratic response Residuals Lack of ﬁt Pure error

3 3 3 10 5 5

866.09 33.03 110.19 44.63 29.99 14.64

288.69 11.00 36.72 4.46 5.99 2.92

64.68 2.46 8.22

7.4 × 10–7 0.12 4 × 10–3

2.04

0.22

Notes: multiple R2: 0.957; adjusted R2: 0.919; F-statistic: 25.13 on 9 and 10 DF; p-value: 0.0000.

Table 2 The design of experiments and experimental responses for P adsorption by ZIF-8. Run order

Standard order

1 2 3 4 5 6 7 8 9 10

2 4 6 14 12 11 3 15 18 19

Coded variable

Response (removal)

X1

X2

X3

observed

predicted

1 1 1 0 0 0 −1 0 0 0

−1 1 −1 0 1.7 −1.7 1 0 0 0

−1 −1 1 1.7 0 0 −1 0 0 0

82.9 84.0 71.5 67.3 78.7 61.8 76.5 75.4 79.3 75.1

82.7 85.3 69.5 70.1 77.5 64.0 78.0 77.1 77.1 77.1

Run order

Standard order

11 12 13 14 15 16 17 18 19 20

13 10 16 7 8 9 5 20 17 1

Coded variable

Response (removal)

X1

X2

X3

observed

predicted

0 1.7 0 −1 1 −1.7 −1 0 0 −1

0 0 0 1 1 0 −1 0 0 −1

−1.7 0 0 1 1 0 1 0 0 −1

88.9 79.9 77.0 71.8 77.3 64.7 59.8 78.8 77.0 68.8

86.8 80.7 77.1 71.4 76.5 64.9 58.0 77.1 77.1 69.1

154

M. Shams et al. / Journal of Molecular Liquids 224 (2016) 151–157

Fig. 2. Experimental P adsorption versus predicted removal efﬁciencies. Fig. 4. The effect of pH on P adsorption. P: 10 mg/L; contact time: 30 min; ZIF-8: 0.5 g/L.

Fig. 3. Contour and 3D response surface plots for P removal. The effect of (a, b) initial P concentration and time, (c, d) ZIF-8 dose and time.

M. Shams et al. / Journal of Molecular Liquids 224 (2016) 151–157

155

100 Removal (%)

35 kHz 80

130 kHz

60

300 rpm 200 rpm

40

100 rpm

20 0 0

10

20

30 40 Time (min)

50

60

70

Fig. 7. Comparison of agitation by mechanical mixer and ultrasound at different conditions on P adsorption. P: 10 mg/L; ZIF-8: 0.5 g/L.

Fig. 5. The effect of coexisting ions on P adsorption. P: 10 mg/L, contact time: 30 min. ZIF-8: 0.5 g/L.

3.3. Effect of the operating variables on P adsorption Graphical representation of the model developed in the present work (Eq. (3)) is shown in Fig. 3(a–d). Contour plots and 3D response surface plots were used to visualize the independent variables effect to the response. These plots provide facile method in ﬁnding the optimum points at which the desired response occurred. P removal as a function of initial P concentration and time shown in Fig. 3(a, b). As expected, P adsorption decreased with increasing initial P concentration. This is expected based on previously reports [21,22]. The higher P concentrations in the solution hinders the P ions to adsorb onto ZIF-8. Fig. 3(c, d) shows the P adsorption increased by ZIF-8 to reach an optimum value. The initial increasing is a consequence of higher available adsorption sites at higher adsorbent doses. Increasing the adsorbent dose higher than optimum value decreased removal efﬁciency. This could attributed to the increasing pH after ZIF-8 added to solution, especially when higher doses were used.

3.4. Effect of pH The effect of pH on P adsorption in the range of 2–12 was evaluated separately in a set of batch experiments. As can be seen in Fig. 4, the P removal increased with decrease in initial solution pH and the maximum adsorption occurred at pH 2.8. This can be explained on the basis of ZIF-8 surface charge. Zero point of charge (pHzpc) of ZIF-8 reported around 9.5 in the literature [22].

(a)

(b)

Fig. 6. ZIF-8 dispersion by agitation under (a) mechanical mixer and (b) ultrasound after 15 min.

The charge of the adsorbent surface getting more positive with decreasing the solution pH and P adsorption increased as a result of higher electrostatic interaction. The decrease in P removal efﬁciency could also attributed to the adsorption competition between P− and OH– ions especially in higher pH. The decrease in percent adsorption by increasing the pH was also reported in the literature on adsorption of arsenic and phthalic acid [23]. Although P adsorption increases under acidic condition, decreasing the solution pH below 2.6 cause a sharp reduction in P adsorption. The sharp fall in P removal could attributed to the instability of ZIF-8 under strong acidic conditions. When pH of the solution adjusted below around 2.6, the solution turbidity disappears gradually due to the dissolution of ZIF-8.

3.5. Effect of competing ions The adsorption of P on ZIF-8 in presence of coexisting anions was studied using different initial concentrations of sulfate, chloride, nitrate, carbonate, and bicarbonate. As shown in Fig. 5, the inﬂuence of the applied anions other than carbonate is negligible for the selected concentrations. The high interference of carbonate on P adsorption could be attributed to changes occurred pH of the solution. In the presence of 200 and 400 mg/L carbonate, the pH level of the solution increased from 5.7 to 8.5 and 9, respectively. The surface charge of ZIF-8 become more negative at higher pH and this is led to a static repulsion of PO34 and HPO24 −. Previous reports also indicated that the carbonate ions can adsorbed by zinc and interfere with the adsorption of other negatively charged ions [21].

3.6. Effect of agitation speed and sonication To study the effect of mixing speed and sonication on P adsorption, the kinetic experiments were carried out under mechanical stirring at 100, 200, and 300 rpm and in ultrasonic bath at 35 and 130 kHz at constant temperature. Fig. 6(a, b) compared samples agitated by conventional mixer and sonication for the same time. As can be seen, the shear forces produced during the microbubble collapsing provide sufﬁcient energy for breaking adsorbent agglomerates and generating a well dispersive mixing system. A comparison of agitation by mechanical mixer and ultrasound at different conditions on P removal efﬁciency is shown in Fig. 7. It can be deduced from Fig. 7 that the increasing turbulence at higher mixing degrees decreases the boundary layer thickness of solid–liquid phase and causes a higher mass transfer rate. According to the Fig. 7, it can also deduced that the equilibrium time can decreased signiﬁcantly when the solution irradiated with a frequency of 35 kHz. Sonication with operating frequency of 130 kHz, on the other hand, was not effective for mixing purpose in the studied time.

156

M. Shams et al. / Journal of Molecular Liquids 224 (2016) 151–157

Table 5 Isotherm models and the parameters of ﬁtted models. Isotherm

Formula

Linear form 1 1 ¼ qm Ce þ qmb

Plot

Parameter

Ce qe

qmax (mg/g) KL (L/mg) R2 KF (mg/g(L/mg)1/n) n R2 kt (L/mg) B1 R2 qmax (mg/g) β R2

Langmuir

qe ¼

Freundlich

qe = KFC1/n e

Log qe = log KF + 1n log Ce

logqe vs. logCe

Temkin

; ln ðkT C e Þ qe ¼ RT b

qe = B1 ln .kt + B1 ln Ce

qe vs. lnCe

Dubinin–Radushkevich

qe =qmexp. (−βε2)

lnqe = lnqm − βϵ2

qe vs. ε2

Ce/qe

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0

(a)

0

Ln qe

Ce qe

qm bC e 1þbC e

10

30

Ce

40

(b)

Adsorption isotherms are important part of a sorption study that describe the pollutants interaction with the adsorbents. Equilibrium relationships also provide information about the adsorbents capacities and their surface properties, which is necessary for designing real sorption systems [34]. Langmuir, Freundlich, Dubinin–Radushkevich, and Temkin models were used to simulate the equilibrium data obtained from P adsorption onto ZIF-8 after 12 h agitation. Table 5 listed the ﬁtted isotherm models, their linear forms, and the values of parameters obtained by ﬁtting the experimental data with the models and shown in Fig. 8(a–d). As it seen, the value of R2 is higher for Langmuir model than the other models. The adsorption theory of Langmuir model is based on the kinetic principles and it proposes the monolayer surface adsorption on the ideal solid with deﬁnite localized sites that are energetically identical. As seen in Table 5, the maximum monolayer adsorption capacity of ZIF-8 for P was 38.22 mg/g in the present work. 3.8. Kinetic modeling

-2

0

4

Ln Ce2

4

(c)

3.5 3 Ln qe

38.22 0.94 0.99 640.3 3.57 0.93 0.59 5.86 0.97 31.1 5.07 0.87

3.7. Equilibrium study

20

4 3.5 3 2.5 2 1.5 1 0.5 0

vs. Ce

2.5 2 1.5

Kinetic studies provide required information on selecting optimum operating conditions in real batch sorption systems. In the present work, the experimental data were ﬁtted to three of the most widely used kinetic models as listed in Table 6 and shown in Fig. 9(a–c). Table 7 shows the ﬁtted experimental data with the pseudo-ﬁrstorder, pseudo-second-order, and intraparticle diffusion kinetic models. As seen, pseudo-second-order kinetic model with higher R2 values showed better compliance with the experimental data, suggesting the adsorption to be controlled by chemosorption.

1 4. Conclusion

0.5 0 0

10

20 2

50 (d)

qe

40 30 20 10 0 -2

-1

0

1 Ln Ce

2

3

4

Fig. 8. Fitting the experimental data with (a) Langmuir, (b) Freundlich, (c) Dubinin– Radushkevich, and (d) Temkin models.

The cubic ZIF-8 synthesized and applied for P removal from aqueous solutions. According to central composite design, a quadratic model was developed for prediction of P removal by ZIF-8. The prediction model suggest that the P removal increased by ZIF-8 dose and decreased with increasing initial P concentration. P removal also increased with decrease in initial solution pH and the maximum adsorption occurred around pH 2.8. Below the optimum pH, P concentration increased sharply due to the dissolution of ZIF-8 in strong acidic condition. Carbonate ions, due to increasing pH, showed a high level of interference on P adsorption. A higher mass transfer rate and removal efﬁciency was obtained when the solution agitated at higher mechanical speed and when the solution agitated by sonication with a frequency of 35 kHz. Present study revealed that the adsorption were ﬁtted well by Langmuir model with a maximum monolayer adsorption capacity of 38.22 mg/g. Pseudo-second-order kinetic model described the experimental data well, suggesting the adsorption to be controlled by chemosorption. Present report suggest ZIF-8 as an efﬁcient and fast removal sorbent for P from aqueous solutions.

M. Shams et al. / Journal of Molecular Liquids 224 (2016) 151–157

157

Table 6 Kinetic models ﬁtted with the experimental data. Kinetic model

Formula

Plot

Pseudo-ﬁrst-order kinetic model

Logðqe −qt Þ ¼

Pseudo-second-order kinetic model

t qt

Intraparticle diffusion kinetic model

qt = kp . t0.5 + c

log(qe − qt) vs. t

k₁ ; logqe − 2:303 :t

t qt

1 1 ¼ k₂qе 2 þ qе :t

vs. t

qt vs. t0.5

Acknowledgment

(a)

The authors are grateful to express their appreciate for the help and support of the laboratory staff members of environmental health department of Tehran University of Medical Sciences.

References

(b)

(c)

Fig. 9. Fitting the experimental data with (a) pseudo-ﬁrst-order, (b) pseudo-second-order, and (c) intraparticle diffusion kinetic models.

[1] Z. Zhu, H. Zeng, Y. Zhu, F. Yang, H. Zhu, H. Qin, W. Wei, Sep. Purif. Technol. 117 (2013) 124–130. [2] M. Shams, S. Dobaradaran, B. Ramavandi, M. Qasemi, M. Afsharnia, Fresenius Environ. Bull. 22 (2013) 722–726. [3] J. Ye, X. Cong, P. Zhang, G. Zeng, E. Hoffmann, Y. Wu, H. Zhang, W. Fang, J. Mol. Liq. 216 (2016) 35–41. [4] W. Gu, Q. Xie, C. Qi, L. Zhao, D. Wu, Powder Technol. 301 (2016) 723–729. [5] C. Han, Z. Wang, W. Yang, Q. Wu, H. Yang, X. Xue, Ecol. Eng. 89 (2016) 1–6. [6] S. Yang, P. Jin, X. Wang, Q. Zhang, X. Chen, Chem. Eng. J. 292 (2016) 246–254. [7] S. Kaneko, K. Nakajima, J. Water Pollut. Control Fed. 60 (1988) 1239–1244. [8] D. Bian, S. Ai, J. Liu, Y. Zuo, X. Tian, J. Environ. Sci. 23 (2011) S106–S109. [9] Y.I. Seo, K.H. Hong, S.H. Kim, D. Chang, K.H. Lee, Y.D. Kim, J. Ind. Eng. Chem. 19 (2013) 744–747. [10] R.J. Seviour, T. Mino, M. Onuki, FEMS Microbiol. Rev. 27 (2003) 99–127. [11] M. Shams, M. Qasemi, S. Dobaradaran, A.H. Mahvi, Fresenius Environ. Bull. 22 (2013) 2604–2609. [12] M.E. Hailian Li, M. O'Keeffe, O.M. Yaghi, Nature 402 (1999) 276–279. [13] H.C. Zhou, J.R. Long, O.M. Yaghi, Chem. Rev. 112 (2012) 673–674. [14] F. Tan, M. Liu, K. Li, Y. Wang, J. Wang, X. Guo, G. Zhang, C. Song, Chem. Eng. J. 281 (2015) 360–367. [15] M. Massoudinejad, M. Ghaderpoori, A. Shahsavani, M.M. Amini, J. Mol. Liq. 221 (2016) 279–286. [16] Y. Pan, W. Liu, Y. Zhao, C. Wang, Z. Lai, J. Membr. Sci. 493 (2015) 88–96. [17] S. Wang, Y. Fan, X. Jia, Chem. Eng. J. 256 (2014) 14–22. [18] F. Cacho-Bailo, B. Seoane, C. Téllez, J. Coronas, J. Membr. Sci. 464 (2014) 119–126. [19] E.L. Bustamante, J. Fernández, J.M. Zamaro, J. Colloid Interface Sci. 424 (2014) 37–43. [20] J. Duan, Y. Pan, F. Pacheco, E. Litwiller, Z. Lai, I. Pinnau, J. Membr. Sci. 476 (2015) 303–310. [21] M. Jian, B. Liu, G. Zhang, R. Liu, X. Zhang, J. Colloid Interface Sci. 465 (2015) 67–76. [22] B. Liu, M. Jian, R. Liu, J. Yao, X. Zhang, J. Colloid Interface Sci. 481 (2015) 358–366. [23] N.A. Khan, B.K. Jung, Z. Hasan, S.H. Jhung, J. Hazard. Mater. 282 (2015) 194–200. [24] B.K. Jung, J.W. Jun, Z. Hasan, S.H. Jhung, Chem. Eng. J. 267 (2015) 9–15. [25] S. Zinatloo-Ajabshir, M. Salavati-Niasari, J. Mol. Liq. 216 (2016) 545–551. [26] G. Ameta, A.K. Pathak, C. Ameta, R. Ameta, P.B. Punjabi, J. Mol. Liq. 211 (2015) 934–937. [27] S. Agarwal, I. Tyagi, V.K. Gupta, M.H. Dehghani, A. Bagheri, K. Yetilmezsoy, A. Amrane, B. Heibati, S. Rodriguez-Couto, J. Mol. Liq 221 (2016) 1237–1242. [28] S.R. Pouran, A. Bayrami, A.R.A. Aziz, W.M.A.W. Daud, M.S. Shafeeyan, J. Mol. Liq 222 (2016) 1076–1084. [29] H. Mazaheri, M. Ghaedi, A. Asfaram, S. Hajati, J. Mol. Liq. 219 (2016) 667–676. [30] M. Rajabi, S. Asemipour, B. Barﬁ, M.R. Jamali, M. Behzad, J. Mol. Liq. 194 (2014) 166–171. [31] R. Jayakumar, M. Rajasimman, C. Karthikeyan, Ecotoxicol. Environ. Saf. 121 (2015) 199–210. [32] M. He, J. Yao, Q. Liu, K. Wang, F. Chen, H. Wang, Microporous Mesoporous Mater. 184 (2014) 55–60. [33] P.F. de Sales, Z.M. Magriotis, M.A.L.S. Rossi, R.F. Resende, C.A. Nunes, J. Environ. Manag. 151 (2015) 144–152. [34] K.Y. Foo, B.H. Hameed, Chem. Eng. J. 156 (2010) 2–10.

Table 7 Constants obtained from kinetic models for P adsorption onto ZIF-8. Pseudo-ﬁrst order

Pseudo-second order

C0 [mg/L]

qe, exp. [mg/g]

qe,cal [mg/g]

K1 [min

5 10 15 20

9.52 18.64 29.36 35.76

3.8 10.59 20.91 27.61

0.06 0.08 0.07 0.054

−1

]

2

R

qe,cal [mg/g]

K2 [min

0.96 0.98 0.97 0.98

9.78 19.45 32.37 37.65

0.04 0.02 0.005 0.003

Intraparticle diffusion −1

]

R

Kp [mg/g min−0.5]

R2

0.99 0.99 0.99 0.99

0.52 1.42 3.11 3.83

0.80 0.86 0.89 0.96

2