Adsorptive removal of Congo red from aqueous solution using zeolitic imidazolate framework–67

Accepted Manuscript Title: ADSORPTIVE REMOVAL OF CONGO RED FROM AQUEOUS SOLUTION USING ZEOLITIC IMIDAZOLATE FRAMEWORK–67 Authors: Nguyen Thi Thanh Tu, Tran Vinh Thien, Pham Dinh Du, Vo Thi Thanh Chau, Tran Xuan Mau, Dinh Quang Khieu PII: DOI: Reference:

S2213-3437(18)30152-0 https://doi.org/10.1016/j.jece.2018.03.031 JECE 2272

To appear in: Received date: Revised date: Accepted date:

12-1-2018 12-3-2018 13-3-2018

Please cite this article as: Nguyen Thi Thanh Tu, Tran Vinh Thien, Pham Dinh Du, Vo Thi Thanh Chau, Tran Xuan Mau, Dinh Quang Khieu, ADSORPTIVE REMOVAL OF CONGO RED FROM AQUEOUS SOLUTION USING ZEOLITIC IMIDAZOLATE FRAMEWORK–67, Journal of Environmental Chemical Engineering https://doi.org/10.1016/j.jece.2018.03.031 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ADSORPTIVE REMOVAL OF CONGO RED FROM AQUEOUS SOLUTION USING ZEOLITIC IMIDAZOLATE FRAMEWORK–67 Nguyen Thi Thanh Tu1,2, Tran Vinh Thien3, Pham Dinh Du4, Vo Thi Thanh Chau5, Tran Xuan Mau1, Dinh Quang Khieu1 1 2 3

Enviromental Technology Center (ENTEC), 700000, Vietnam

Faculty of Natural Sciences, Phu Yen University, 620000 Vietnam 5

IP T

Faculty of Natural Sciences, Thu Dau Mot University,590000, Vietnam Industrial University of Ho Chi Minh City, 570000, Vietnam

SC R

4

University of Sciences, Hue University, 530000, Vietnam

Graphical abstract SO3

H 2N 

Electrostatic interaction

R NH2

O3S

+ + + + + + + + + + + + + + + + + +

NH

N

R O3S

N

+ + + + + +

NH2

R

pp stacking interaction

(CGR)

R'

Co N

N

N

N

ZIF-67

SO3

H 2N 

Coordination interaction

R

NH2

O3S Co

N

N

N

M

N

A

N



N

SO3

H 2N + + + + + + + +N +Co +N + + + + + + + +

Co 2+

U

+ + + + + +

D

Highlights

The piecewise linear regression combined with Akaike’s Information Criteria is a useful statistical tool for analyzing kinetics and isotherms models.



ZIF–67 exhibits very high adsorption for some other dyes such as Congo red, Rhodamine B and methylene blue.



ZIF–67 is expected to become one of the most promising adsorbents to remove dyes from water.

A

CC

EP

TE



Abstract In the present paper, the Congo red dye (CGR) adsorption onto zeolitic imidazolate framework–67 (ZIF–67) is demonstrated. ZIF–67 was synthesized using the microwave method. The obtained ZIF–67 was characterized by means of X–ray diffraction (XRD), scanning electron microscope (SEM), thermal gravity analysis (TG), and X–ray photoelectron spectroscopy (XPS). ZIF–67 was employed to adsorb CGR from aqueous solutions. The first-

IP T

order and second-order kinetic model, and Weber’s intra-particle-diffusion and Boyd’s filmdiffusion model were utilized to study the adsorption kinetics. The equilibrium adsorption data at different temperatures were interpreted using the Langmuir and Freundlich models. The thermodynamics from 301 K to 331 K was also addressed. ZIF–67 exhibited excellent adsorption in terms of favourable thermodynamics, ultra-high adsorption and superb fast kinetics. The adsorption capacity of ZIF–67 for CGR could be as high as 714.3 mg·g−1 at 301 K, which could be improved at higher temperatures. In addition, ZIF–67 also exhibited very high adsorption for some other dyes such as rhodamine B and methylene blue. ZIF–67 was stable and could also be conveniently reused for several times. These peculiarities facilitate ZIF–67 to become one of the most promising adsorbents to remove dyes from water.

SC R

Keywords: ZIF–67, Congo red, Boyd’s film-diffusion model; Weber’s intra-particle diffusion model

U

I. Introduction

N

Congo red is the sodium salt of 3,3′-([1,1′-biphenyl]-4,4′-diyl)bis(4-aminonaphthalene– 1-sulfonic acid). It is a secondary diazo dye. Congo red (henceforth CGR) is water-soluble,

A

yielding a red colloidal solution [1]. It has a strong, though apparently noncovalent, affinity

M

for cellulose fibres. CGR has been widely used in the textile industry, wood pulp, and paper production. Discharging CGR into water resources even in a small amount can affect the aquatic

D

life and food web. CGR is expected to metabolize to benzidine, which has been reported to be carcinogenic and mutagenic for aquatic organisms [2]. Therefore, finding efficient and friendly

TE

techniques to remove dyes in general and CGR in particular from wastewater has become an urgent and important issue. Many common physical chemistry methods have been developed

EP

for this purpose, including biological oxidation [3], adsorption [1, 4, 5], photocatalysis [6, 7], chemical coagulation [8, 9], ion exchange [10, 11]. Among these methods, adsorption is

CC

considered to be one of the most promising techniques due to high efficiency, low cost, and operation simplicity. So far, various natural and synthetic adsorbents such as bagasse fly ash

A

and commercial activated carbon [1], coal-based mesoporous activated carbon [4], chitosan hydrogel beads impregnated with nonionic or anionic surfactants [5], hierarchical Ni(OH)2 and NiO nanosheets [12], chitosan hydrogel beads impregnated with cetyl trimethyl ammonium bromide [13], chitosan hydrogel beads impregnated with carbon nanotubes [14], spinel CoFe2O4 [15], natural zeolites modified with N,N-dimethyl dehydroabietylamine oxide [16], high surface area activated carbon [17] have been developed to remove CGR. However, many of these adsorbents have certain deficiencies, including low adsorption capability and low

recyclability. Hence, developing new adsorbent materials with a high surface area and excellent adsorption capability is actually still an important question for basic study and practical application [18]. Zeolitic imidazolate frameworks (ZIFs), a class of metal-organic frameworks (MOFs), are formed from tetrahedral metal ions (e.g., Zn, Co) bridged by imidazolate. ZIFs exhibit unique and highly desirable properties from both the zeolites and the metal-organic frameworks. Hence, ZIFs have attracted considerable interest in many applications such as

IP T

heterogeneous catalysis [19, 20]; adsorption [21, 22]; and chemical sensing [23, 24]. ZIF–67 (Co(mim)2, mim = 2-methylimidazole) is the most representative ZIF. It is formed by bridging

SC R

2 methyl imidazolate anions with a cobalt cation with a sodalite-type topology. Recently, ZIF– 67 has received much attention and widely been studied in several fields of adsorption. ZIF–67 has been used in the adsorptive removal of rhodamine B (RhB), anionic methyl orange (MO), cationic methylene blue (MB) [25], anionic dye acid blue 40 [26], malachite green [21], and Cr

U

(IV) [22]. These studies found out that ZIF–67 can be highly stable in the aqueous phase and

N

possesses very high adsorption capacities. With these peculiarities, ZIF–67 can be a relevant adsorbent for removing CGR from water. However, to the best of our knowledge, few studies

A

have been carried out to study the capability of ZIF–67 to remove CGR from aqueous solutions.

M

Due to the porous structure of MOFs, the studies of dye adsorption have focused on diffusion kinetics. The Weber’s intra-particle-diffusion model and Boyd’s film-diffusion model

D

have been widely used to interpret the kinetics data. However, Weber or Boyd’s linear plot,

TE

especially Weber’s plot, often composes the multi-linearity in nature. In general, the researchers visually determine how many linear segments there are, then choose points to divide the plot into these linear segments [21, 27]. The uncertainty in deciding the linear segments possibly

EP

entrains the uncertainties in determining their slopes and intercepts. This results in uncertainties in estimating the diffusion coefficients. Therefore, one cannot determine exactly whether film

CC

the diffusion, intra-particle diffusion, or the transition between them takes place in the sorption process. Malash et al. [28] proposed using a statistical approach of the piecewise linear

A

regression for the analysis of experimental adsorption data by means of the diffusion models. This method, which allows the determination of each linear segment and the number of linear segments as well, would avoid the subjective decisions that may arise with the graphical approach. Despite these facts, the application of this statistical analysis for adsorption kinetics data is still limited [29, 30]. In the thermodynamic adsorption studies, the thermodynamic parameters including Gibbs free energy (∆G0), enthalpy (∆H0), and entropy (∆S0) have been determined from the expression

SC R

IP T

∆G0 = –RT·lnKeq and ∆G0 = ∆H0 – T·∆S0 at different temperatures. The Langmuir equilibrium constant, KL, in the Langmuir isotherm is usually used as Keq to calculate the thermodynamic parameters. Since the unit for the term RT is also J·mol−1, the equilibrium constant, Keq, must be dimensionless. Therefore, Milonjić and followed by Zhou et al. [31, 32], Tran et al. [33] proposed the way to transform the dimension of KL, e.g., mg–1·L or mol–1·L into dimensionless by multiplying by appropriate coefficients. While Liu [34] reported that the use of the Langmuir equilibrium constant with units of litres per mol for the calculation of ∆G0 and subsequent determination of ∆H0 and ∆S0 of adsorption would be acceptable for a dilute solution of charged adsorbates or for a solution of uncharged adsorbates at any concentrations. Moreover, Keq for the thermodynamic calculation has been derived differently, for example, as the Langmuir equilibrium constant with dimension (mol·L–1 or mg·L–1) [35, 36]; the distribution constant [37–39], and so on. As a result, values of ∆G0, ∆H0, and ∆S0 of adsorption reported in the literature are very confusing. Few articles summarize and compare these differences [33, 34]. In the present paper, the adsorption of the CGR dye onto ZIF–67 has been performed. The Weber’s intra-particle-diffusion model, Boyd’s film-diffusion model, and pseudo-

U

first/second-order kinetic models were used to interpret the mechanism of adsorption. Akaike’s

N

information criteria (AIC) and the piecewise linear regression method were used to analyze the

A

kinetics data. The adsorption isotherms and kinetics of CGR adsorption onto ZIF–67 were

M

studied at different temperatures. The thermodynamic data (∆H0, ∆S0, ∆G0) calculated from

TE

2. EXPERIMENTAL

D

different equilibrium constants were addressed.

2.1. Materials and characterization Cobalt(II)

nitrate

hexahydrate

((Co(NO3)2·6H2O,

Daejung,

Korea)

and

2-

EP

methylimidazole (C4H6N2, Sigma-Aldrich, U.S.A., 99%) were utilized in this paper. Congo red (C32H22N6(SO3)2Na2, (Molecular weight = 696.66) was supplied by Dindni, Nashik, India. The

A

CC

molecular structure of CGR is shown in Scheme1.

Scheme 1. Structure of the CGR molecule

The powder X-ray diffraction (XRD) patterns were performed by means of D8 Advance, Bruker (Germany) with CuK radiation ( = 1.5406 Å). The morphology of ZIF–67 was observed with a scanning electron microscope (SEM) using SEM JMS–5300LV (Japan). Nitrogen adsorption/desorption isotherms were carried out using a Micromeritics 2020 volumetric adsorption analyzer system (U.S.A.). The thermal behaviour of the obtained ZIF– 67 was analyzed by means of thermal analysis (TG–DTA) using Labsys TG Setaram (France). X-ray photoelectron spectroscopy (XPS) was performed using a Shimadzu Kratos

IP T

AXISULTRA DLD spectrometer, Japan. Visible spectrophotometry was measured using a

A

N

U

SC R

Lambda 25 Spectrophotometer – Perkin Elmer (Singapore) at max of CGR dye (600 nm).

2.2. Preparation of ZIF–67

M

Scheme 2. Microwave device for ZIF–67 synthesis

D

The microwave synthesis for ZIF–67 was carried out using a hand-made microwave

TE

device fabricated from a domestic microwave oven (Sharp R–203VN–M) with the power of 250 W as shown in Scheme 2. Typically, an exact amount of imidazole (16 mmol) and Co(NO3).6H2O (4 mmol) was dissolved in 200 mL of a mixture of ethanol: distilled water:

EP

DMF (1: 1: 1 in w/w). The resulting mixture was loaded into a 250 mL Erlenmeyer flask and placed in the hand-made microwave device. The microwave irradiation time was 30 min; then,

CC

the solid was collected by using centrifugation and washed with DMF three times and dried at 100 °C for 24 hours. Since the microwave was attached with a reflux condenser, the synthesis

A

was carried out under atmospheric pressure. The temperature was 70–90 °C depending on the irradiation time. 2.3. Kinetic studies Adsorption kinetics studies were performed in a batch process. A 3-litre plastic beaker equipped with a stainless-steel flat-blade impeller driven by an electric motor was used to stir the dye solution. ZIF–67 (0.2 g) was vigorously stirred with 2000 mL of the CGR solution in

the beaker at a fixed temperature. 5 mL of solution was drawn at pre-set times through a tap, and the solid was separated using a centrifuge. The residual dye concentrations were determined using UV–Vis spectrophotometry. The dye adsorption capacity of ZIF–67 was calculated using Eq. 1 𝑞𝑡 =

𝑉 · (𝐶0 − 𝐶𝑡 ) 𝑚

(1)

where qt is the dye adsorption capacity (mg·g–1) at time t (min); C0 is the initial dye

volume of the dye solution (L); and m is the mass (g) of ZIF–67.

IP T

concentration (mg·L−1); Ct is the concentration (mg·L−1) of the dye solution at time t; V is the

𝑞𝑒 =

SC R

The adsorption capacity, qe (mg·g–1), at equilibrium is calculated according to Eq. 2 𝑉 · (𝐶0 − 𝐶𝑒 ) 𝑚

(2)

where Ce is the dye concentration at equilibrium. Other parameters are described above.

U

The formal adsorption kinetics was interpreted using the pseudo-first-order and pseudo-

N

second-order kinetic models.

A

The pseudo-first-order kinetic model in the non-linear form is the Lagergren equation [40] written as

M

𝑞𝑡 = 𝑞𝑒 · (1 − 𝑒 −𝑘1 ·𝑡 )

(3)

defined previously.

D

where k1 is the rate constant of the pseudo-first-order model (min–1). The other parameters are

TE

The pseudo-second-order kinetic model in the non-linear form is expressed as follows

EP

[41]

𝑞𝑡 = 𝑞𝑒

𝑞𝑒 · 𝑘2 · 𝑡 1 + 𝑞𝑒 · 𝑘2 · 𝑡

(4)

CC

where k2 is the rate constant of the pseudo-second-order kinetic model (g·mg–1·min–1). k1, k2, and qe were calculated by means of the non-linear regression method using the

A

Solver function in Microsoft Excel. The kinetics of diffusion was studied by means Weber’s intra-particle-diffusion and

Boyd’s film-diffusion models. Weber’s intra-particle-diffusion model is described in Eq. 5 [42] 𝑞𝑡 = 𝑘𝑖𝑑 . 𝑡1/2 + 𝐼

(5)

where kid is the intra-particle-diffusion rate constant (mg·g–1·min–0.5); I is the intercept, which reflects the boundary-layer effect on the adsorption capacity.

If the linear plot of qt versus t1/2 provides a zero intercept, the intra-particle diffusion is the rate-limiting step. The Boyd’s film-diffusion model is expressed in Eq. 6 [43] ln(1 − 𝐹) = 𝑘𝑓𝑑 · 𝑡

(6)

where F is the fractional attainment of equilibrium at different times (F = qt/qe); and kfd is the film-diffusion rate constant (min–1).

IP T

A linear plot of ln(1−F) versus t with the zero intercept would suggest that the kinetics of the sorption process is controlled by intra-particle diffusion. If the plot is nonlinear or linear but does not pass through the origin, then it is concluded that the film diffusion or the chemical

SC R

reaction controls the adsorption rate [43, 44].

To determine whether the lines cross the origin, we statistically compare the intercept with the zero value by determining its 95 % confidence interval. If the intercept is statistically

U

different from zero, the interval does not contain zero and vice versa. This process is done using

N

SPSS–20 (Statistical Package for the Social Sciences–20).

In spite of their apparent simplicity, the plot of Weber’s model often composes many

A

linear segments rather than just one. In the present study, the analysis of the multi-linearity in

M

Weber’s plot was performed by using the piecewise linear regression suggested by Malash et al. [28].

D

In this method, the experimental data could be fixed for a one-, two-, three- or four-linear-

TE

segment line in Weber’s model as follows One-linear-segment line: Y = B + A·X (two parameters)

EP

Two-linear-segment line: Y = B + A·X +C·(X–D)·SIGN(X–D) (four parameters) Three-linear-segment line: Y = B + A·X + C·(X–D)·SIGN (X – D) + E·(X – F)·SIGN (X

CC

– F) (six parameters)

Four-linear-segment line: Y = B + A·X + C·(X–D)·SIGN (X – D) + E·(X–F)·SIGN (X –

A

F) + G·(X – H)·SIGN(X – H) (eight parameters) where the values of A, B, C, D, E, F, G, and H are estimated by means of nonlinear regression. D, F, and H called breakpoints are the boundaries between the segments. The Microsoft Excel “SIGN” function is defined as follows 1 if 𝑋 > 𝑎 𝑆𝐼𝐺𝑁(𝑋 − 𝑎) = ( 0 if 𝑋 = 𝑎) −1 if 𝑋 < 𝑎

(7)

The example for the three-linear-segment equation is expressed as in Eq. 8 𝐴 + 𝐶 · 𝐷 + 𝐸 · 𝐹 + 𝑋 · (𝐵 − 𝐶 − 𝐸) if 𝑋 < 𝐷 𝐴 + 𝐸 · 𝐹 + 𝑋 · (𝐵 − 𝐸) if 𝑋 = 𝐷 𝑌 = 𝐴 − 𝐶 · 𝐷 + 𝐸 · 𝐹 + 𝑋 · (𝐵 + 𝐶 − 𝐸) if 𝐷 < 𝑋 < 𝐹 𝐴 − 𝐶 · 𝐷 + 𝑋 · (𝐵 + 𝐶) if 𝑋 = 𝐹 ( 𝐴 − 𝐶 · 𝐷 − 𝐸 · 𝐹 + 𝑋 · (𝐵 + 𝐶 + 𝐸) if 𝑋 > 𝐹 )

(8)

Then, the linear equation of the first segment is y = b1·x + a1 where b1 = B – C – E, and a1 = A + C·D + E·F.

IP T

The linear equation of the second segment is y = b2·x + a2 where b2 = B + C – E, and a2 = A – C·D + E·F.

SC R

The linear equation of the third segment is y = b3·x + a3 where b3 = B + C + E, and a3 = A – C·D – E·F.

The nonlinear regression determines the model’s parameters by the least square method. They are calculated by minimizing the sum of squared deviations, SSES, by means of the

U

numerical optimization techniques using the Solver function in Microsoft Excel. The function

N

for minimization is 𝑁

(9)

M

1

A

𝑆𝑆𝐸𝑆 = ∑(𝑦𝑒𝑥𝑝 − 𝑦𝑒𝑠𝑡 )2

where yexp is the experimental response, and yest is the response estimated from the model.

TE

D

The determination coefficient R2 is obtained according to the expression 𝑅2 =

1 − 𝑆𝑆𝐸𝑆 𝑆𝑆𝐸𝑇

(10)

2 where SSET is the total sum of squares, equal to (∑𝑁 1 (𝑦𝑒𝑥𝑝 − 𝑦𝑚𝑒𝑎𝑛 ) (where ymean is the mean

EP

value of y).

When we compare two models, the one with more parameters will always fit the

CC

experimental data better (lower sum of squared deviations) than the one with fewer parameters. There are two common ways to compare the two models, i.e., F-test and AIC (Akaike’s

A

information criteria). The F-test is only employed to related models, while the AIC is applied to both non related and related models. AIC is defined as in Eq. 11 𝐴𝐼𝐶 = 𝑁 · ln(𝑆𝑆𝐸𝑆 /𝑁) + 2 · 𝑁𝑝

(11)

where N is the number of data points; Np is the number of parameters fit by the regression.

The second-order AIC (or corrected AIC, AICc) is applied when N is small compared to Np. AICc is only computed when N is at least two units greater than Np. AICc [29, 30, 45] is calculated for each model from Eq. 12 𝐴𝐼𝐶𝑐 = 𝑁 · ln(𝑆𝑆𝐸𝑇 /𝑁) + 2 · 𝑁𝑝 +

2 · 𝑁𝑝 · (𝑁𝑝 + 1) 𝑁 − 𝑁𝑝 − 1

(12)

AICc (or AIC) evaluates how well the data support each model. The value of AICc can be

IP T

positive or negative. The model with the lowest AICc score is the most likely correct. 2.4. Equilibrium studies

For adsorption equilibrium experiments, an amount of 0.014; 0.017; 0.02; 0.026; 0.03;

SC R

and 0.035 mg of ZIF–67 was added to six 250-mL Erlenmeyer flasks filled with 200 mL CGR solutions (80 mg·L–1) at neutral pH (ca. 7). The flasks were sealed and shaken for 24 h to confirm saturation at four temperatures (301 K, 311 K, 321 K, and 331 K). Afterwards, the solids were separated by means of centrifugation. The concentration of CGR remaining in the

U

supernatant solutions was determined using the method mentioned above.

N

The equilibrium data were analyzed according to the Freundlich and Langmuir isotherm

A

models.

M

Langmuir isotherm: The Langmuir equation is valid for the monolayer adsorption onto the surface. It could be expressed as [46, 47] .

D

1 1 1 = + 𝑞𝑒 𝑞𝑚𝑜𝑚 · 𝐾𝐿 · 𝐶𝑒 𝑞𝑚𝑜𝑚

(13)

TE

where qmom is the maximum monolayer capacity amount (mg·g–1); KL is the Langmuir equilibrium constant (L·mg–1); and the others are described above.

EP

The Langmuir parameters were obtained from the slope and intercept of the linear plot of 1/qe versus 1/Ce.

CC

The essential characteristics of the Langmuir isotherm can be expressed through a

A

dimensionless constant separation factor, RL, which is expressed as 𝑅𝐿 =

1 1 + 𝐶0 · 𝐾𝐿

(14)

where the value of RL indicates the type of isotherm: unfavourable (RL > 1), linear (RL = 1), favourable (0 < RL < 1) or irreversible (RL = 0) [48]. Freundlich isotherm: The Freundlich equation is an empirical relation based on the adsorption of the adsorbates onto a heterogeneous surface. It is expressed as [47, 49, 50].

ln𝑞𝑒 =

1 · ln𝐶𝑒 + ln𝐾𝐹 𝑛

(15)

where KF is the Freundlich constant, which is a measure of adsorption capacity; n is an empirical parameter. A large value of n indicates that the surface is heterogeneous. For values in the range 1 < n < 10, the adsorption is favourable. The values of n between 2 and 10 imply good sorption processes; whereas, 1 < n < 2 shows that adsorption capacity is only slightly suppressed at lower

IP T

equilibrium concentrations [51]. The Freundlich parameters were obtained from the slope and intercept of the linear plot of lnqe versus lnCe.

SC R

The linear correlation is significant if its p-value is less than the significance level (α = 0.05). In the present paper, the linear correlation was conducted by means of SPSS–20. 2.5. Thermodynamic studies

U

The experimental procedure was conducted the same way as the study of adsorption kinetics. However, the temperature of the process was fixed at 301 K, 311 K, 321 K, and

N

331 K. The activation energy, Ea, was determined using the Arrhenius equation [52]. 𝐸𝑎

A

𝑘 = 𝐴 · 𝑒 −𝑅·𝑇

(16)

M

where k is the rate constant equal to the rate constant k2 in the pseudo-second-order kinetic

absolute temperature (K).

D

equation; A is the frequency factor; R is the gas constant (8.315 J·mol–1·K–1); and T is the

ln 𝑘 = −

𝐸𝑎 + ln 𝐴 𝑅·𝑇

(17)

EP

TE

Taking the natural logarithm of both sides of Eq. 16, one obtains

By linear plotting of lnk versus 1/T, Ea could be obtained from the slope (–Ea/R).

CC

The thermodynamic parameters of activation including enthalpy (∆𝐻 # ), entropy ∆𝑆 # , and Gibbs free energy ∆𝐺 # of activation for CGR adsorption kinetics were obtained by applying

A

Eyring equation [53, 54]. The Eyring equation in its thermodynamic version is as follows 𝑘𝑏 · 𝑇 −∆𝐺# 𝑘𝑏 · 𝑇 ∆𝑆# −∆𝐻 # 𝑘= · 𝑒 𝑅·𝑇 = · 𝑒 𝑅 · 𝑒 𝑅·𝑇 ℎ ℎ

(18)

where k is the rate constant equal to the rate constant k2 in the pseudo-second-order equation; kb (1.3807×10–23 J·K–1) is the Boltzmann constant; and h (6.621×10–34 J·s) is the Planck constant.

Taking the natural logarithm of both sides of Eq.18, one can have the Eyring equation in the linear form as 𝑘 𝑘𝑏 ∆𝑆 # ∆𝐻 # ln = ln + − 𝑇 ℎ 𝑅 𝑅·𝑇 𝑘

(19)

1

By plotting of ln 𝑇 versus 𝑇 , one can obtain ∆𝑆 # and ∆𝐻 # from the slope ( intercept (ln

𝑘𝑏 ℎ

+

∆𝑆 # 𝑅

∆𝐻 # 𝑅

) and y-

).

IP T

The Gibbs free energy of activation can be obtained by using Eq. 20 ∆𝐺 # = ∆𝐻 # − 𝑇 · ∆𝑆 #

(20)

SC R

In order to assess if the adsorption is spontaneous or not, the thermodynamic parameters of adsorption are evaluated. The standard Gibbs free energy of adsorption (ΔG0) is given by Eq. 21 [52].

(21)

U

∆𝐺 0 = ∆𝐻 0 − 𝑇 · ∆𝑆 0

respectively.

A

∆𝐺 0 is given by van’t Hoff’s equation

N

where ∆𝐺 0 , ∆𝐻 0 and ∆𝑆 0 are the standard Gibbs free energy, enthalpy, and entropy,

M

∆𝐺 0 = −𝑅𝑇 · ln 𝐾

(22)

where K is the thermodynamic equilibrium constant, and the others are described above.

TE

D

By replacing Eq. 22 to Eq. 21, one obtains ln 𝐾 = −

∆𝐻 0 ∆𝑆 0 + 𝑅·𝑇 𝑅

(23)

1

EP

The plot of ln 𝐾 versus yields a straight line that allows the calculation of ∆H0 and ∆S0 𝑇

from the respective slope and intercept of Eq. 23.

CC

3. RESULTS AND DISCUSSION 3.1. Characterization of ZIF–67

A

Figure 1a shows the XRD pattern of a ZIF–67 sample, where all the well-defined

diffraction peaks match well with the standard simulation XRD pattern of ZIF–67 calculated from reported structural data [55], indicating that the obtained sample has a pure phase with high crystallinity. The SEM observation is presented in Figure 1b. It can also be seen that these ZIF–67 particles have good uniformity, well-defined facets, and sharp edges and corners. Its morphology consists of crystals of well-defined truncated rhombic dodecahedrons with a size

of around 700–1000 nm. The porous properties of ZIF–67 were studied by means of nitrogen adsorption/desorption isotherms as shown in Figure 1c. The isotherm curves belong to type I according to the IUPAC classification; the curves are characteristic for micro-porous materials. The specific surface area calculated using the BET model is 1388 m2·g–1, which is higher or compatible with that reported in the literature [56, 57]. The thermal analysis is a useful tool for studying the thermal stability. Figure 1d illustrates the TG diagram of ZIF–67 in the nitrogen atmosphere. One can see that ZIF–67 first

IP T

loses its weight from room temperature to 300 °C (4.11 %) due to the escape of water molecules as well as gas molecules from the cavities. Subsequently, a major weight loss of ZIF–67 (27.2

SC R

%) occurs in the range of 250–600 °C, which could be attributed to the decomposition of the

(011)

ligands.

N

U 20

30

40

50

M

10

A

(022) (013) (222) (114) (233) (244) (134) (044) (334) (235)

(002)

(112)

Intensity (arb.)

a

60

D

2 theta / degree

550

450 400

EP

500

d -27.178 % -1.068mg

80

60

-41.169 % -1.618mg

40

Adsorption

Desorption

20

350

0.0

-4.114 % -0.162mg

100

TG / %

TE

c

CC

3 -1 Adsorption volume / cm . g

600

0.2

0.4 0.6 0.8 Relative pressure / P/P0

1.0

0

100

200

300

400

500

600

700

800

900 1000 1100

Temperature / 0C)

A

Figure 1. a) XRD pattern; b) SEM observation; c) nitrogen adsorption/desorption isotherms; d) TG diagram of ZIF–67 The XPS spectrum of the ZIF–67 sample was studied to further confirm the chemical

composition and chemical state of the elements. The survey spectrum in Figure 2a shows that the ZIF–67 sample is composed of C, O, N, and Co [58–61]. The high-resolution XPS spectrum of C1s in Figure 2b can be fitted to three peaks at 284.9 eV, 286.4 eV, and 289.0 eV. The main

peak at 284.9 eV corresponds to the C‒C bond in the imidazole ring. The peak at 286.4 eV and 289.0 eV might be characteristic of carbon in the C–N and C=N bonds, respectively. The existence of an only peak of N2 (299.1 eV) that corresponds to N of the 2-methylimidazole ligand thoroughly coordinates to Co [58] (Figure 2c). The core level XPS spectrum of Co2p represents two peaks at 779.72 eV and 794.72eV, which can be attributed to Co 2p3/2 and Co

EP

TE

D

M

A

N

U

SC R

IP T

2p1/2 of Co2+ in ZIF–67, respectively [58] (Figure 2d).

CC

Figure 2. a) XPS spectrum of ZIF–67; b) XPS core level spectrum of C1s; c) XPS core level spectrum of N1s; d) XPS core level spectrum of Co2p The stability of adsorbents in different acidic media is important in terms of application.

A

In the present study, ZIF–67 materials were immersed in water at different pHs for 24 hours. The stability of ZIF–67 was tested by means of XRD measurement. Figure 3 shows that the XRD pattern at pH =1 is different from the original pattern of ZIF–67, indicating that ZIF–67 is unstable at this low pH. Meanwhile, the XRD patterns of the samples at pH = 2 ÷ 12 remain unchanged. Therefore, the crystallinity of ZIF–67 materials retains well, which proves that ZIF– 67 is stable in aqueous solutions in this pH range.

20000 cps

Intensity (arb.)

pH = 12 pH = 8 pH = 6 pH = 4 pH = 3 pH = 2 pH = 1 20

30 40 2 theta / degree

50

60

SC R

10

IP T

Initial ZIF-67

Figure 3. XRD patterns of ZIF–67 in water with different pHs

U

3.2. Congo red adsorption onto ZIF–67 3.2.1. Adsorption kinetics

N

The adsorption kinetics of CGR onto ZIF–67 with different initial concentrations is

M

A

presented in Figure 4.

650

D

600

500 450

30 mg L-1 50 mg L-1

400 350

70 mg L-1 80 mg L-1

300 0

10

20

30

40

50

60

70

80

Time / min

A

CC

EP

TE

qe / mg. g-1

550

Figure 4. Dependence of adsorption capacity on contact time at different initial concentrations; Adsorption conditions: initial CGR concentration = 30 ÷ 100 mg·L–1; mass of adsorbent = 0.2 g; volume of solution = 2000 mL; stirring speed = 300 rpm The initial concentration provides important information about the driving force that overcomes all the mass transfer resistance of CGR between the aqueous solution and the surface

of ZIF–67. As a result, a high initial CGR concentration will promote the sorption process. The equilibrium adsorption capacity increases from 300 mg·g–1 to 632 mg·g–1 as the initial CGR concentration increases from 30 mg·L–1 to 80 mg·L–1. Figure 4 reveals that the adsorption of CGR is very fast in the first few minutes (0–5 min). In particular, the adsorption takes place almost instantaneously at low concentrations (30 mg·L–1). The time required to reach the adsorption equilibrium between ZIF–67 and CGR at higher initial concentrations is less than 40 min.

IP T

The pseudo-first-order and pseudo-second-order kinetic models were applied to interpret the kinetics data. Since the experimental points and the number of parameters for the two

SC R

models are the same, the determination coefficients (R2) were used to compare the models. The kinetic parameters of the models and R2 are listed in Table 1.

Table 1. Parameters of the pseudo-first-order and pseudo-second-order kinetic models

k1

qe,cal

qe,exp

–1

–1

k2

300.7

300.0

0.97

0.57

0.017

444.9

446.0

0.93

613.0

0.45

0.016

610.6

613.0

0.90

632.0

0.44

0.013

628.4

632.0

0.94

299.7

300.0

50

2.321

436.4

446.0

70

2.598

604.3

80

2.330

0.77

TE

M

A

4.445

–1

R2

0.213

30

–1

qe,exp (mg·g–1)

(mg·g )

–1

qe,cal (mg·g )

(mg·g )

621.3

R2

(mg ·g ·min )

(min )

D

–1

Second-order kinetic model

U

(mg·L–1)

First-order kinetic model

N

Concentration

It was found that the experimental data of CGR adsorption onto ZIF–67 do not follow

EP

the pseudo-first-order model due to the low values of R2 (0.44–0.77). Whereas, the experimental results fit the pseudo-second-order model well: the straight lines with high values of R2 (0.90–

CC

0.97) were obtained and qe,cal agrees with qe,exp. The pseudo-second-order kinetic rate coefficient decreases from 0.213 mg–1·g·min–1 to 0.013 mg–1·g·min–1 when the initial CGR concentration increases from to 30 mg·L–1 to 80 mg·L–1 (Table 1). This behaviour is also observed in other

A

studies [62, 63]. This is because increasing the dye concentration might reduce the diffusion of dye molecules in the boundary layer and hence enhance the diffusion in the pores of the solid [63].

25 mg.g-1

One segment

qe / mg . g-1

Two segments Three segments

1

2

3

4 5 1/2 1/2 t / min

6

7

8

SC R

0

IP T

Four segments

Figure 5. Plots of piecewise linear regression for one, two, three and four segments based on

U

Weber’s intra-particle diffusion model; Adsorption conditions: initial CGR concentration = 80

N

mg·L–1; mass of adsorbent = 0.2 g; volume of solution = 2000 mL; stirring speed = 300 rpm

A

The intra-particle diffusion is expected due to the porous structure of ZIF–67. Therefore, Weber’s plots are drawn by putting the CGR adsorption capacity, qt , against the square root of

M

time, t1/2. However, previous studies by various researchers showed that these plots are multilinear in nature [28, 41, 63]. Hence, the piecewise linear regressions were used to analyze the

D

data. Figure 5 illustrates the experimental data and piecewise linear regression lines with the

TE

initial concentration of 80 mg·L–1 for ZIF–67. From the figure, the experimental points are close to the regression lines for the three- and four-linear-segment models. Since increasing the

EP

number of linear segments leads to an increase in the number of regression parameters, a decrease in SSE or an increase in R2 naturally follows. For this reason, SSE or R2 are only used

CC

to compare the models with the same number of parameters and solely cannot be used to decide the goodness of fit for models with different number of parameters. Only the experimental data for the initial concentration of 50, 70 and 80 mg·L–1 were used for the analysis using the multi-

A

linear-segment model because at low CGR concentration (30 mg·L–1), the number of experimental data is not sufficient due to the rapid and instantaneous adsorption (Figure 4). It can be observed from Table 2 that when the number of parameters increases, the SSE decreases or R2 increases, and R2 is approximately equal to 1 for the three- or four-segment model. It is hard to decide whether the three-segment or four-segment model is a better description of the experimental data based on SSE or R2. Therefore, in this study, the compatibility of the models

was assessed by using AICc, and the smaller of which indicates that the model is more compatible with experimental data. Among the four models, it can be seen that the three-linearsegment model fits the data best because it has the smallest AIC values (Table 2). Table 2. Comparison of piecewise linear regression for one-, two-, three- and four-linear segments using AIC One-linear-segment regression

CCGR (mg·L–1)

Two-linear-segment Regression

Three-linear-segment regression

Four-linear-segment Regression

R2

AIC

SSE

R2

AIC

SSE

R2

AIC

SSE

R2

AIC

50

348.42

0.85

50.1

61.39

0.97

33.1

3.44

0.99

4.3

3.38

0.99

24.9

70

598.76

0.99

62.9

74.55

0.99

36.3

42.83

0.99

37.1

23.68

0.99

42.8

80

1376.02

0.71

95.6

74.40

0.98

37.2

9.15

0.98

–1.7

6.40

0.99

–0.07

SC R

IP T

SSE

U

The data of the three-linear-segment regression for different initial concentrations are presented in Table 3. The intercept of the first segment in the Weber plot is 516.08 with 95 %

N

confidence interval between 352.44 and 679.73 at the initial concentration of 80 mg·L–1. This

A

value of the intercept is significantly different from zero. It means that the line does not pass

M

the origin. The similar behaviours are also observed for the other cases. These results indicate that the intra-particle diffusion is not the only rate-limiting step.

D

In the transport of mass from the liquid phase to the solid phase, the boundary plays the

TE

most significant role in adsorption, and Boyd’s liquid-film-diffusion model may be applied to understand the process.

EP

Table 3. Results of piecewise regression analysis for the three-linear-segment model for ZIF–67 (The values in parentheses are at the 95 % confidence interval) First linear segment

CC

CCGR

A

(mg·L–1)

Second linear segment

Third linear segment

Intercept 1

Slope 1

Intercept 2

Slope 2

Intercept 3

Slope 3

50

366.76 (169.89; 563.63)

32.25

403.80 (398.06; 409.54)

11.09

430.45 (405.64; 455.26)

3.37

70

537.93 (390.00; 685.86)

26.58

546.89 (529.48; 564.31)

19.60

597.91 (595.71; 600.11)

2.24

80

516.08 ( 352.44; 679.73)

51.59

598.04 (596.04; 600.04)

5.50

615.46 (609.94; 620.97)

2.12

Table 4. Parameters of Boyd’s film-diffusion model for adsorption of CGR onto ZIF–67 Film-diffusdion rate constant

CCGR (mg·L–1)

Intercept (95 % confidence interval)

R2

30

1.128

–2.98 (–5.12; –0.83)

0.91

50

0.203

–2.19 (–2.29; –2.09)

0.99

70

0.204

–2.59 ( –2.89; –2.31)

80

0.081

–2.83 (–2.99; –2.66)

IP T

kfd (min−1)

0.96

SC R

0.97

A linear plot of ln(1−F) versus t with the non-zero intercept would suggest that the kinetics of the sorption process is controlled by diffusion through the liquid film surrounding

U

the solid adsorbents. The film-diffusion rate constants gained from the slope and the confidence

N

interval for the intercepts are listed in Table 4. As seen from the table, the rate constant for liquid film diffusion, kfd, decreases from 1.128 min−1 to 0.08 min−1 as the initial concentration

A

increases from 30 mg·L–1 to 80 mg·L–1. It is understandable because the higher the density of

M

molecules on the adsorbent surface, the lower is the diffusion capacity. The curves exhibit linear plots with high determination coefficient values (R2 = 0.91 ÷ 0.99). However, the confident

D

interval does not contain zero. Therefore, both intra-particle diffusion and film diffusion are

TE

participating in controlling the diffusion of the CGR molecules. 3.2.2. Equilibrium adsorption studies

EP

Two common isotherm models: Langmuir and Freundlich were used to analyse the equilibrium data at different temperatures. As can be seen in Table 5 and Figure 6, both models

CC

provide high determination coefficients (R2 = 0.95 ÷ 0.99). In the statistical view, the linear regression between 1/Ce and 1/qe, and lnCe and lnqe do have statistical significance because pvalues are less than 0.05 in all cases [64]. Furthermore, the characteristic parameters (i.e., RL

A

for the Langmuir isotherm and n for the Freundlich isotherm) are favourable with RL in the range of 0 to 1 and the value of n = 3.2 ÷ 3.8. It is concluded that both isotherms well explain the sorption process, implying a monolayer adsorption and the existence of heterogeneous surface in the adsorbents. The maximum adsorption capacity increases with the increase in the temperature indicating that the CGR adsorption has favourable thermodynamics. The adsorption capacity of ZIF–67 for CGR is 714.3 mg·g−1; 769.2 mg·g−1; 833.3 mg·g−1; and 909.1

mg·g−1 at 301 K; 311 K; 321 K and 331 K, respectively. Table 5. Parameters of Langmuir and Freundlich models at different temperatures Langmuir isotherm model

Freundlich isotherm model

Temp. KL (L·mg–1)

R2

p–value

301

714.3

0.272

0.95

0.003

311

769.2

0.324

0.95

0.004

321

833.3

0.353

0.98

0.001

331

909.1

0.461

0.99

< 0.001

KF (L·g–1)

272.601 291.102

3.7

0.99

< 0.001

3.4

0.98

< 0.001

0.97

< 0.001

0.98

< 0.001

3.8

381.213

b

p–value

U

a

R2

3.2

313.822

6.8 0.0026

n

IP T

qmom (mg·g–1)

SC R

(K)

6.7

0.0024

N

6.6

0.0022

lnqe

6.5

A

0.0018

301 K 311 K

0.0014

321 K

0.0012

331 K

0.0010 0.0

0.1

D

M

0.0016

0.2

0.3

TE

1/qe

0.0020

1/Ce

0.4

0.5

6.4 6.3

301 K

6.2

311 K

6.1

321 K

6.0

331 K

0.5

1.0

1.5

2.0

2.5

3.0

3.5

lnCe

CC

EP

Figure 6. Plots of Langmuir isotherm model (a) and Freundlich isotherm model (b) for CGR adsorption onto ZIF–67; Adsorption conditions: initial CGR concentration = 80 mg·L–1; mass of adsorbent = 0.014 ÷ 0.035 g; volume of solution = 200 mL; shaking time = 24 hours In addition, adsorption isotherms were also applied to investigate the maximum

adsorption capacity of methylene blue (MB) and rhodamine B (RhB) onto ZIF–67. The results

A

show that the experimental data follow the Langmuir model. The maximum monolayer adsorption capacity of MB and RhB onto ZIF–67 is 50.5 mg·g–1 and 95.8 mg·g–1, respectively. A comparison of the CGR, MB, and RhB adsorption capacities onto ZIF–67 with other previously reported adsorbents [1, 4, 5, 12–17] is presented in Table 6. It is worth noting that ZIF–67 exhibits a super adsorption for CGR. The adsorption capacity of ZIF–67 for CGR is 2 to 10 times higher than that of most adsorbents announced such as high-surface-area activated

carbon [17], chitosan hydrogel beads impregnated with nonionic or anionic surfactant [5], cobalt ferrite [15], etc. The adsorption capacity for RhB or MB on ZIF–67 is also higher than or compatible with that of other adsorbents [65–69]. It is clear that ZIF–67 possesses an ultrahigh adsorption capacity towards anion dyes (Congo red, rhodamine B) and a cation dye (methylene blue) compared with other adsorbents. Table 6. Adsorption capacity of different adsorbents towards CGR, MB, and RhB at ambient temperature Adsorbents

Adsorption capacity

(m2·g–1)

(mg·g–1) 714.3

The present study

11.8

[1]

0.637

[1]

52–189

[4]

–

182–378

[5]

127–201

39.7–151.7

[12]

CGR

–

352

[13]

CGR

237.8

450.4

[14]

Dye

ZIF–67

CGR*

1388

2

Bagasse fly ash

CGR

168

3

Commercial activated carbon

CGR

390

CGR

370–679

7

8

U

N

A

CGR

M

6

CGR

D

5

Coal-based mesoporous activated carbon Chitosan hydrogel beads impregnated with nonionic or anionic surfactant Hierarchical Ni(OH)2 and NiO nanosheets Chitosan hydrogel beads impregnated with cetyl trimethyl ammonium bromide Chitosan hydrogel beads impregnated with carbon nanotubes

TE

4

SC R

1

IP T

Order

Specific area

References

Spinel CoFe2O4

CGR

N/A

244.5

[15]

10

Natural zeolites modified with N,N–dimethyl dehydroabietylamine oxide

CGR

N/A

69.49

[16]

11

High surface area activated carbon

CGR

2797

512

[17]

50.5

The present study

CC

EP

9

ZIF–67

13

Al–MCM–41

MB

N/A

66.5

[69]

14

Indian Rosewood sawdust

MB

N/A

56.4

[65]

15

Fe3O4/ZIF–8

MB

1068

20.2

[66]

16

ZIF–67

95.8

The present study

17

Activated carbon

307

[67]

A

12

MB**

RhB*** RhB

720

18

Jute stick powder

RhB

32.6

87.7

[68]

19

Al–MCM–41

RhB

625

91

[69]

* CGR: Congo red; ** MB: Methylene blue; *** Rhodamine B

3.2.3. Thermodynamic studies The thermodynamic studies were performed at 301 K, 311 K, 321 K, and 331 K. The adsorption kinetics curves at different temperatures are presented in Figure 7. From the figure,

IP T

the equilibrium adsorption capacity, qeq, increases with the increase in temperature, which indicates that the process is endothermic. The thermodynamic parameters and activation energy

SC R

are presented in Table 7. It is obvious that the pseudo-second-order kinetic mode fits well with the kinetic data in the temperature range of 301 K to 331 K because it provides higher R2 than the pseudo-first-order kinetic mode. Therefore, the adsorption rate constant, k2, was used to

U

calculate the thermodynamic parameters.

N

720 700

A

680

M

640 620 600

D

qe / mg.g-1

660

TE

580

301 K 311 K 321 K 331 K

560 540

20

40

EP

0

60

80

Time / min

CC

Figure 7. Effect of temperature on adsorption of CGR dye onto ZIF–67; Adsorption conditions: initial CGR concentration = 80 mg·L–1; mass of adsorbent = 0.200 g; volume of solution = 2000 mL

A

Table 7. Activation energy and rate constants for CGR adsorption onto ZIF–67

Temp. (K) 301

First-order kinetic equation

Second-order kinetic equation

qeq (mg·g–1)

k1 (min–1)

R2

qeq (mg·g–1)

k2 (mg ·g·min–1)

Ea (kJ·mol–1)

R2

621.5

2.100

0.76

630.5

0.011

16.50

0.96

–1

311

673.5

2.537

0.59

680.6

0.014

0.95

321

698.1

2.752

0.58

704.2

0.017

0.93

331

708.4

2.895

0.54

713.9

0.020

0.92

The activation energy refers to the minimum amount of energy that must be overcome for the adsorption to take place. Ea for ZIF–67 is 16.5 kJ·mol–1. This low activation energy

IP T

(below 42 kJ·mol–1) implies the diffusion-controlled process because the temperature dependence of pore diffusivity is relatively weak, and the diffusion process refers to the movement of the solute to the external surface but not its diffusivity to the microspore surfaces

SC R

in the particle [54, 70]. It means that the rate-limiting step, in this case, involves a physicochemical mechanism but not purely a physical or chemical one.

Table 8. Activation parameters for CGR adsorption onto ZIF–67 H# (kJ·mol–1)

U

S# (J·mol–1·K–1)

N

Temp. (K)

A

301 –240.4 321

84.9 87.3

12.6 89.7 92.2

TE

D

331

M

311

G# (kJ·mol–1)

The activation parameters derived from the Eyring equation were investigated in order to determine whether the sorption process follows an active complex prior to the final product

EP

(Table 8). The determination coefficient (R2 = 0.98, p = 0.01) is high although the Eyring equation is not strictly linear with respect to 1/T [62]. The value of ∆S# is also an indication of

CC

whether or not a reaction has an associative or dissociative mechanism. The entropy values > – 10 J·K–1·mol–1 generally indicate a dissociative mechanism [54]. However, in Table 8, one sees

A

a large negative value of ∆S# (–240.4 J·K–1·mol–1), suggesting that the sorption process exhibits an associate mechanism [54], in which the possibility of an associative chemisorption through the formation of an activated complex between CGR molecules and ZIF–67 takes place [53].

I N U SC R

The relation between Ea and ∆𝐻 # for the reaction in solutions is described in Eq. 24 [52] 𝐸𝑎 = ∆𝐻 # + 𝑅 · 𝑇

(24)

A

The calculated values of ∆𝐻 # from Eq. (24) vary from 13.7 kJ·mol–1 to 14.00 kJ·mol–1. Meanwhile, the obtained value of 12.6 kJ·mol–1 for

M

∆𝐻 # is slightly lower than the calculated values. It means that CGR adsorption onto ZIF–67 is not merely chemisorption but also physisorption. The large positive ∆𝐺 # in the CGR adsorption onto ZIF–67 implies that this reaction requires energy to convert the reactants to the product, and

TE

D

when the energy requirement is satisfied, the reaction takes place.

Table 9. Thermodynamic parameters calculated using different equilibrium constants

CC EP

Parameters based on Kd

Parameters based on Ke

Temp.

(K)

Kd

( L·g–1)

A

301 311

∆H0

(kJ·mol–1)

∆S0

(J·mol–1·K–1)

37.62 58.14

22.9

∆G0 (kJ·mol–1)

∆H0 Ke

–9.1

3.76

–10.5

5.81

107.1

∆S0 –

(kJ·mol 1 )

22.9

(J·mol –1 ·K–1)

Parameters based on KL (Dimension)

∆G0

KL

(kJ·mol–1)

(L·mg–1)

–3.32

0.27

–4.60

0.32

88.2

∆H0

∆S0

∆G0

(kJ·mol –1 )

(J·mol– 1 ·K–1 )

(kJ·mol– 1 )

14.1

Parameters based on KL (Dimensionless) KLx10-7

∆H0

∆S0

∆G0

(kJ·mol–1)

(J·mol–1·K–1 )

(kJ·mol–1)

3.29

1.04

-40.43

2.97

1.22

-42.20

35.6

14.1

180.9

321

75.65

–11.6

7.57

–5.40

0.35

2.78

1.36

-43.85

331

86.39

–12.3

8.64

–5.93

0.46

2.15

1.77

-45.93

The thermodynamic parameters including ∆H0, ∆S0 and ∆G0 were calculated based on Eq. 23. It is obvious that the determination of the equilibrium constant (Keq) is a key towards the correct values of ∆G0, ∆H0, and ∆S0. The dimensional Langmuir equilibrium constant (L·mg–1) (dimensional KL) [38, 39] and dimensionless KL (transformed by multiplying by 103×55.5×MW (MW: The molecular mass of CGR) [31, 33] has been employed to calculate the 𝑞

thermodynamic parameters. The distribution constant 𝐾𝑑 = 𝐶𝑒 (L·g–1) is often used in some 𝑒

studies of adsorption [38]. In other studies of adsorption thermodynamics [71, 72], the 𝐶0 −𝐶𝑒 𝐶𝑒

(dimensionless) is used for the calculation of ∆G0 of

IP T

distribution constant 𝐾𝑒 =

EP

TE

D

M

A

N

U

SC R

adsorption. In this study, the thermodynamic parameters based on different equilibrium constants are listed in Table 9. The sorption process in ZIF–67 is endothermic as indicated by the positive sign of the ΔH0 value. This means that the magnitude of the total energy absorbed by the system due to bond breaking is greater than the magnitude of the total energy released due to bond forming [62]. The positive value of ΔS0 suggests the increase in randomness in the system as the CGR molecules get adsorbed onto ZIF–67. The negative value of ΔG0 means the sorption process is spontaneous, whereas the positive value of ΔG0 means the sorption process is not favourable thermodynamically [52]. As seen from Table 9, the magnitude and sign of ΔH0, ΔS0, and ΔG0 depend significantly on the applied equilibrium constant. The thermodynamic parameters ΔS0, ΔH0, and ΔG0 have the same sign when derived from the constants Kd, Ke, and KL (dimensionless), while KL (L·mg-1) provides the opposite sign for ΔG0. Therefore, the discussion about the spontaneity is diverse, even contradictory. However, from the analysis of isothermal adsorption and thermodynamics, it is obvious that ZIF–67 has a very high adsorption capacity for CGR, and its adsorption capacity is proportional to the temperature; therefore, it can be concluded that the adsorption of CGR onto ZIF–67 is spontaneous. Selecting the equilibrium constant to calculate the thermodynamic parameters is different from each study [31–36, 38]. Which constant is the most appropriate is still controversial. In this study, we found that the equilibrium constant Kd, Ke or dimensionless KL is more appropriate for computation than dimensional KL (mg·L–1).

CC

3.2.4. Effect of pH and proposed mechanism The ionization of pollutants and the ionic state of the functional groups on the surface of

the adsorbent depend significantly on the pH of the solution. Figure 8 illustrates the effect of

A

pH on the CGR adsorption capacity. As seen in the figure, the adsorption capacity of CGR onto ZIF–67 decreases gradually with the increase in pH 6–8.5 and decreases significantly afterwards. pK1 and pK2 of CGR in the solution are 3 and 5, respectively [73], and CGR exists as the negative species in the solution above pH = 5 and positive or neutral species below this value. The pHPZC of ZIF–67 determined by means of the pH drift method is around 9 (the inset of Figure 8). When the solution pH is below the pHPZC, the ZIF–67 surface has a positive charge,

while at high pH (> 9.0) it has a negative charge. With an increase in pH, the positive charge of ZIF–67 surface decreases resulting in the decline of adsorption. At pH higher than pHpzc, the adsorption capacity decreases significantly due to the negatively charged surface repulsing the

700

6

680

4 Delta pH

660

620

2

pHPZC = 9 0 4

6

8

-2

600

10

12

pH

SC R

qe / mg.g-1

640

IP T

negative species of CGR via electrostatic repulsion.

580 560 540

500 7

8

9

10 pH

11

12

13

14

N

6

U

520

A

Figure 8. a) Effect of pH on CGR adsorption capacity; b) pHZPC (zero point charge) determined

M

using pH drift method; Adsorption conditions: initial CGR concentration = 80 mg·L–1; mass of

D

adsorbent = 0.020 g, volume of solution = 200 mL; shaking time = 24 hours This electrostatic interaction mechanism is similar to the adsorption mechanism of

TE

phthalic acid [74] and p-arsanilic acid on ZIF–8 [75]. The π−π stacking interaction between the aromatic rings of CGR and the aromatic imidazole rings of the ZIF–67 framework is also

EP

responsible for the CGR efficient adsorption. In practice, this interaction is used to explain the high adsorption capacity of adsorbents containing the sp2 hybridization of carbon with

CC

adsorbates containing aromatic rings in their molecular structure [21, 29]. In addition, the coordination of nitrogen and oxygen atoms in CGR molecules to the Co2+ ions in the framework of ZIF–67 is also thought to contribute to CGR adsorption. From these arguments, a possible

A

mechanism of CGR adsorption onto ZIF–67 is suggested (Scheme 3).

SO3

H 2N 

Electrostatic interaction

R NH2

O3S

+ + + + + + + + + + + + + + + + + + + + + + + +

SO3

Co 2+ N

NH

+ + + + + + + +N +Co +N + + + + + + + + N



R O3S

NH2

R

pp stacking interaction

(CGR)

N

+ + + + + +

R'

Co N

N

N

N

ZIF-67

H 2N 

SO3

R

O3S

NH2

SC R

Coordination interaction

IP T

H2N

Co

N

N

N

N

U

Scheme 3. The proposed mechanisms of the CGR onto ZIF–67

N

3.2.5. Reusability of ZIF–67

A

To evaluate the reusability of ZIF–67 for the removal CGR, the used adsorbents were

M

regenerated by washing with a solution of ethanol 30 % (in volume) in an ultrasonic bath for several times until a clear water was obtained and then drying for 24 hours at 100 °C. The

D

regenerated adsorbents were used to adsorb CGR again. The adsorption capacity of the

TE

regenerated ZIF–67 is shown in Figure 9a. The CGR adsorption ability decreases gradually with the increase in the desorption cycles. However, the regenerated adsorbents still perform

EP

well. After three cycles of desorption, the regenerated adsorbents provide 90 % of their adsorption capacity for CGR compared with the initial material. The XRD patterns of the

CC

adsorbents after the third recycling seem to be unchanged (Figure 9b), indicating that the

A

adsorbents are stable under the working conditions.

b

20000 cps

a 600

Initial ZIF-67

Intensity (arb.)

qe / mg g-1

500 400 300 200

first cycle second cycle

100

third cycle

0 first cycle

second cycle

third cycle

10

20

30

40

50

60

IP T

Initial

2 theta / degree

Figure 9. a) Effect of regeneration cycles of ZIF–67 on CGR adsorption; b) XRD patterns of

SC R

ZIF–67 after three cycles; Adsorption conditions: volume of solution = 1000 mL; initial CGR concentration = 80 mg·L–1; adsorbent/VCGR = 0.2 g/2000 mL; shaking time = 10 hours

4. CONCLUSIONS

U

The kinetic, isotherm and thermodynamic studies for the adsorption of Congo red onto

N

zeolite imidazole framewok–67 were demonstrated. The piecewise linear regression combined

A

with Akaike’s Information Criteria is a useful statistical tool for analyzing the kinetics and isotherms models. The analysis shows that the pseudo-second-order kinetic mechanism is more

M

likely, and both intra-particle diffusion and film diffusion control the diffusion of the CGR molecules. The experimental isothermal data fit well with both Langmuir and Freundlich model

D

in the temperature range of 301–331 K. The congo red adsorption capacity of ZIF–67 is

TE

714.3 mg·g−1 at room temperature. In addition, ZIF–67 also possesses a high adsorption capacity with methylene blue and rhodamine B. ZIF–67 is stable in the aqueous solution and

EP

exhibits a superb adsorption in terms of favourable thermodynamics, very high adsorption and very fast kinetics. Therefore, ZIF–67 might serve as one of the effective and promising MOFs

CC

adsorbents for removing dyes from aqueous solutions.

A

Declarations of interest: none

A

CC

EP

TE

D

M

A

N

U

SC R

IP T

References 1. I. D. Mall, V. C. Srivastava, N. K. Agarwal, and I. M. Mishra, Chemosphere, 2005, 61, 492–501. 2. R. Han, D. Ding, Y. Xu, W. Zou, Y. Wang, Y. Li, and L. Zou, Bioresource technology, 2008, 99, 2938–2946. 3. S. Ghoreishi and R. Haghighi, Chemical Engineering Journal, 2003, 95, 163–169. 4. E. Lorenc-Grabowska and G. Gryglewicz, Dyes and pigments, 2007, 74, 34–40. 5. S. Chatterjee, D. S. Lee, M. W. Lee, and S. H. Woo, Bioresource Technology, 2009, 100, 3862–3868. 6. U. Akpan and B. Hameed, Journal of hazardous materials, 2009, 170, 520–529. 7. K. Rajeshwar, M. Osugi, W. Chanmanee, C. Chenthamarakshan, M. V. B. Zanoni, P. Kajitvichyanukul, and R. Krishnan-Ayer, Journal of photochemistry and photobiology C: photochemistry reviews, 2008, 9, 171–192. 8. A. K. Verma, R. R. Dash, and P. Bhunia, Journal of environmental management, 2012, 93, 154–168. 9. F. El-Gohary and A. Tawfik, Desalination, 2009, 249, 1159–1164. 10. C. -H. Liu, J. -S. Wu, H. -C. Chiu, S. -Y. Suen, and K. H. Chu, Water Research, 2007, 41, 1491–1500. 11. J. -S. Wu, C. -H. Liu, K. H. Chu, and S. -Y. Suen, Journal of membrane science, 2008, 309, 239–245. 12. B. Cheng, Y. Le, W. Cai, and J. Yu, Journal of Hazardous Materials, 2011, 185, 889– 897. 13. S. Chatterjee, D. S. Lee, M. W. Lee, and S. H. Woo, Bioresource Technology, 2009, 100, 2803–2809. 14. S. Chatterjee, M. W. Lee, and S. H. Woo, Bioresource Technology, 2010, 101, 1800– 1806. 15. L. Wang, J. Li, Y. Wang, L. Zhao, and Q. Jiang, Chemical Engineering Journal, 2012, 181, 72–79. 16. S. Liu, Y. Ding, P. Li, K. Diao, X. Tan, F. Lei, Y. Zhan, Q. Li, B. Huang, and Z. Huang, Chemical Engineering Journal, 2014, 248, 135–144. 17. J. Li, D. H. Ng, P. Song, C. Kong, Y. Song, and P. Yang, Biomass and Bioenergy, 2015, 75, 189–200. 18. C. Lei, X. Zhu, B. Zhu, C. Jiang, Y. Le, and J. Yu, Journal of hazardous materials, 2017, 321, 801–811. 19. L. Yang, L. Yu, G. Diao, M. Sun, G. Cheng, and S. Chen, Journal of Molecular Catalysis A: Chemical, 2014, 392, 278–283. 20. T. Truong, T. M. Hoang, C. K. Nguyen, Q. T. Huynh, and N. T. Phan, RSC Advances, 2015, 5, 24769–24776. 21. K. -Y. A. Lin and H. -A. Chang, Chemosphere, 2015, 139, 624–631. 22. X. Li, X. Gao, L. Ai, and J. Jiang, Chemical Engineering Journal, 2015, 274, 238-246. 23. H. Zhao, X. Li, W. Li, P. Wang, S. Chen, and X. Quan, RSC Advances, 2014, 4, 36444– 36450. 24. L. Xiao, H. Xu, S. Zhou, T. Song, H. Wang, S. Li, W. Gan, and Q. Yuan, Electrochimica Acta, 2014, 143, 143–151. 25. Y. Li, K. Zhou, M. He, and J. Yao, Microporous and Mesoporous Materials, 2016, 234, 287–292. 26. X. -Z. Kang, Z. -W. Song, Q. Shi, and J. -X. Dong, Asian Journal of Chemistry, 2013, 25, 8324.

A

CC

EP

TE

D

M

A

N

U

SC R

IP T

27. N. L. Torad, M. Hu, S. Ishihara, H. Sukegawa, A. A. Belik, M. Imura, K. Ariga, Y. Sakka, and Y. Yamauchi, small, 2014, 10, 2096–2107. 28. G. F. Malash and M. I. El-Khaiary, Chemical Engineering Journal, 2010, 163, 256– 263. 29. M. Thi Thanh, T. Vinh Thien, V. Thi Thanh Chau, P. Dinh Du, N. Phi Hung, and D. Quang Khieu, Journal of Chemistry, 2017, Article ID 5045973. 30. B. H. Dang Son, V. Quang Mai, D. Xuan Du, N. Hai Phong, and D. Quang Khieu, Journal of Chemistry, 2016, Article ID 8685437. 31. S. K. Milonjic’, Journal of the Serbian chemical society, 2007, 72, 1363-1367. 32. X. Zhou and Xin Zhou, Chemical Engineering Communications, 2014, 201, 14591467. 33. H. N. Tran, S. J. You and H. P. Chao, Journal of Environmental Chemical Engineering, 2016, 4, 2671-2682. 34. Y. Liu, Journal of Chemical & Engineering Data, 2009, 54, 1981–1985. 35. Y. -S. Ho and A. E. Ofomaja, Biochemical Engineering Journal, 2006, 30, 117–123. 36. C. -H. Weng, C. -Z. Tsai, S. -H. Chu, and Y. C. Sharma, Separation and Purification Technology, 2007, 54, 187–197. 37. J. Romero-Gonzalez, J. Peralta-Videa, E. Rodrıguez, S. Ramirez, and J. GardeaTorresdey, The Journal of chemical thermodynamics, 2005, 37, 343–347. 38. R. Sheha and A. El-Zahhar, Journal of hazardous materials, 2008, 150, 795–803. 39. M. F. Sawalha, J. R. Peralta-Videa, J. Romero-González, and J. L. Gardea-Torresdey, Journal of Colloid and Interface Science, 2006, 300, 100–104. 40. R. -L. Tseng, F. -C. Wu, and R. -S. Juang, Journal of the Taiwan Institute of Chemical Engineers, 2010, 41, 661–669. 41. M. A. Al-Ghouti, M. A. Khraisheh, M. N. Ahmad, and S. Allen, Journal of Hazardous Materials, 2009, 165, 589–598. 42. W. J. Weber and J. C. Morris, Journal of the Sanitary Engineering Division, 1963, 89, 31–60. 43. G. E. Boyd, A. W. Adamson, and L. S. M. Jr, J. Am. Chem. Soc., 1947, 69, 2836–2848. 44. S. H. Jeong and K. Park, International journal of pharmaceutics, 2008, 361, 26–32. 45. H. Motulsky and A. Christopoulos, Fitting models to biological data using linear and non linear regression, A practical guide to curve fitting, in Graph Pad Softwave Inc Graph Pad Softwave Inc., Sandiego CA, www.graphpad.com. 2003. 46. I. Langmuir, Journal of the Franklin Institute, 1917, 183, 102–105. 47. M. Fathi, A. Asfaram, and A. Farhangi, Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2015, 135, 364–372. 48. A. Asfaram, M. Ghaedi, S. Hajati, A. Goudarzi, and E. A. Dil, Ultrasonics sonochemistry, 2017, 34, 1–12. 49. S. Agarwal, I. Tyagi, V. Gupta, A. Bagheri, M. Ghaedi, A. Asfaram, S. Hajati, and A. Bazrafshan, Journal of Environmental Chemical Engineering, 2016, 4, 1769–1779. 50. H. M. F. Freundlich, J. Phys.Chem, 1906, 57, 1100-1107. 51. S. -C. Ma, J. -L. Zhang, D. -H. Sun, and G. -X. Liu, Applied Surface Science, 2015, 359, 48–54. 52. P. Atkins and J.d. Paula, Physical Chemistry. 9th Edition 9th ed., W. H. Freeman and Company. 53. T. Anirudhan and P. Radhakrishnan, The Journal of Chemical Thermodynamics, 2008, 40, 702–709. 54. K. G. Scheckel and D. L. Sparks, Soil Science Society of America Journal, 2001, 65, 719–728.

A

CC

EP

TE

D

M

A

N

U

SC R

IP T

55. R. Banerjee, A. Phan, B. Wang, C. Knobler, H. Furukawa, M. O'keeffe, and O. M. Yaghi, Science, 2008, 319, 939–943. 56. H. Yang, X. -W. He, F. Wang, Y. Kang, and J. Zhang, Journal of Materials Chemistry, 2012, 22, 21849–21851. 57. Y. Pan, Z. Li, Z. Zhang, X. -S. Tong, H. Li, C. -Z. Jia, B. Liu, C. -Y. Sun, L. -Y. Yang, and G. -J. Chen, Journal of environmental management, 2016, 169, 167–173. 58. G. Yu, J. Sun, F. Muhammad, P. Wang, and G. Zhu, RSC Advances, 2014, 4, 38804– 38811. 59. J. Pan, Y. Mao, H. Gao, Q. Xiong, F. Qiu, Tao. Zhang and X. Nui, Carbohydrate polymers, 2016, 143, 212-222. 60. Y. Zhang, Y. Shen, Y. Chen, Y. Yan, J. Pan, Q. Xiong, W. Shi and L. Yu, Chemical Engineering Journal, 2016, 294, 222-235. 61. X. Niu, Q. Xiong, J. Pan, X. Li, W. Zhang, F. Qiu, Y. Yan, Fuel, 2017, 190, 174-181. 62. C. -C. Kan, M. C. Aganon, C. M. Futalan, and M. L. P. Dalida, Journal of Environmental Sciences, 2013, 25, 1483–1491. 63. Z. Al-Qodah, W. Lafi, Z. Al-Anber, M. Al-Shannag, and A. Harahsheh, Desalination, 2007, 217, 212–224. 64. A. Field, Discovering statistics using SPSS. 2009: Sage publications. 65. V. K. Garg, M. Amita, R. Kumar, and R. Gupta, Dyes and pigments, 2004, 63, 243– 250. 66. J. Zheng, C. Cheng, W. -J. Fang, C. Chen, R. -W. Yan, H. -X. Huai, and C. -C. Wang, CrystEngComm, 2014, 16, 3960–3964. 67. L. Li, S. Liu, and T. Zhu, Journal of Environmental Sciences, 2010, 22, 1273–1280. 68. G. C. Panda, S. K. Das, and A. K. Guha, Journal of Hazardous Materials, 2009, 164, 374–379. 69. S. Eftekhari, A. Habibi-Yangjeh, and S. Sohrabnezhad, Journal of hazardous materials, 2010, 178, 349–355. 70. D. L. Sparks, Advances in Agronomy, 1986, 38, 231–266. 71. R. Qadeer, Journal of Zhejiang University. Science. B, 2005, 6, 353-356. 72. R. Han, H. Li, Y. Li, J. Zhang, H. Xiao, and J. Shi, Journal of hazardous materials, 2006, 137, 1569–1576. 73. R. W. Sabnis, Handbook of acid-base indicators. 2007: CRC Press. 74. N. A. Khan, B.K. Jung, Z. Hasan, and S. H. Jhung, Journal of hazardous materials, 2015, 282, 194–200. 75. B. K. Jung, J. W. Jun, Z. Hasan, and S. H. Jhung, Chemical Engineering Journal, 2015, 267, 9–15.