A. A. Bazrafshana,
S. Hajati*b and
M. Ghaedi*a
aChemistry Department, Yasouj University, Yasouj 75918-74831, Iran. E-mail: m_ghaedi@mail.yu.ac.ir; m_ghaedi@yahoo.com; Fax: +98 74 33223048; Tel: +98 74 33223048
bDepartment of Physics, Yasouj University, Yasouj 75918-74831, Iran. E-mail: hajati@mail.yu.ac.ir; Fax: +98 74 33223048; Tel: +98 74 33223048
First published on 1st September 2015
Zn(OH)2 nanoparticles (Zn(OH)2-NPs) were sonochemically synthesized. A small amount of Zn(OH)2-NPs was loaded onto activated carbon with a weight ratio of 1:
10 followed by characterization using scanning electron microscopy (SEM), X-ray diffraction (XRD), Fourier transform infrared spectroscopy (FTIR) and diffuse reflectance spectroscopy (DRS). Both activated carbon (AC) and Zn(OH)2 nanoparticle-loaded activated carbon (Zn(OH)2-NP-AC) as safe, green and cost-effective adsorbents were used for the removal of malachite green (MG). Response surface methodology as a cost-effective and time-saving approach was applied to model and optimize dye removal versus adsorbent mass, pH, initial dye concentration and sonication time as well as to investigate the possible interaction among these variables. Zn(OH)2-NP-AC, even at a small nanoparticle loading (with a weight ratio of 1
:
10), was found to be more efficient than AC. For the Zn(OH)2-NP-AC, the optimum values were found to be 0.019 g, 4.5, 20 mg L−1 and 8.6 min for the adsorbent mass, pH, initial dye concentration and sonication time, respectively. The experimental equilibrium data were then fitted to the conventional isotherm models such as Langmuir, Freundlich, Temkin and Dubinin–Radushkevich. The Langmuir isotherm was found to be the best model for the explanation of the experimental data. The adsorption monolayer capacity of Zn(OH)2-NP-AC was obtained to be 74.63 mg g−1, which is comparable to published reports. Adsorption kinetics was studied at various initial MG concentrations, which showed that the adsorption of MG follows the pseudo-second-order rate equation, in addition to the intraparticle diffusion model. The adsorbent was shown to be highly regenerable over several iterations. The short-time adsorption process, high adsorption capacity and good regenerability of the safe, green and cost-effective Zn(OH)2-NP-AC make it advantageous and promising for wastewater treatment.
Biosorption,4 oxidation,5 photocatalytic degradation6 and adsorption7–11 have been commonly used for such purposes.12,13 Adsorption achieved through the use of highly efficient, cost-effective, non-toxic and easily available adsorbents is the most popular approach.14 Simple-to-design activated carbon-based materials with and without loading satisfy these criteria with favorable capability for the adsorption of high amounts of wide-ranging pollutants.15,16 Adsorption processes using activated carbon (AC) with improved surface areas have been widely proposed and used for the removal of both organic and inorganic pollutants from aqueous effluents.17,18 These adsorbents (particularly on nanoscales) possess high capacities for wastewater treatment.19,20 Ghaedi et al.7–11,18,19 widely studied dye adsorption using nanomaterial-loaded activated carbon. They have shown that the nanomaterial-loaded AC is more efficient than the AC for dye removal because of the improved surface area and number of surface reactive atoms. Metallic nanostructures, such as NiSe and ZnSe nanoparticles,9 Pt nanoparticles,10 ZnS:Cu nanoparticles18 and Cd(OH)2 nanowires,19 have functional groups on their surface and thus are easily loaded onto OH and COOH present at the surface of AC, which causes high improvement in the surface area and adsorbate–adsorbent interaction. On the other hand, ultrasound irradiation significantly enhances the mass transfer from adsorbate to adsorbent in addition to their mixing due to the ultrasound-assisted formation of cavitations. This is in turn because of rapid and local changes in the pressure. The bubbles would undergo violent compression when subjected to high pressure, which can generate an intense shockwave.
In this study, Zn(OH)2 nanoparticles (Zn(OH)2-NPs) were sonochemically synthesized. A small amount of Zn(OH)2-NPs was loaded onto AC with a weight ratio of 1:
10. SEM, XRD, FTIR and diffuse reflectance spectroscopy (DRS) were used for their characterization. AC and Zn(OH)2 nanoparticle-loaded AC (Zn(OH)2-NP-AC) were used as safe, green and cost-effective adsorbents for MG removal. Response surface methodology (RSM) as a cost-effective and time-saving method was applied to model and optimize MG removal, whereas adsorbent mass, pH, initial MG concentration and sonication time were considered as variables. RSM was also applied to investigate the possible interaction between the variables involved. The optimal conditions were predicted, the isotherm and kinetics of the adsorption were also studied, and the regenerability of (Zn(OH)2-NPs) was investigated.
Factors | Level | Low (−1) | Central (0) | High (+) | Star point α = 2.0 | |
---|---|---|---|---|---|---|
−α | +α | |||||
(X1) MG concentration (mg L−1) | 10 | 15 | 20 | 5 | 25 | |
(X2) adsorbent dosage (g) | 0.0060 | 0.0105 | 0.0150 | 0.0015 | 0.0195 | |
(X3) pH | 3.5 | 5 | 6.5 | 2 | 8 | |
(X4) sonication time (min) | 3.0 | 5.5 | 8.0 | 0.5 | 10.5 |
Run | (X1) | (X2) | (X3) | (X4) | R%MG |
---|---|---|---|---|---|
1 | 15 | 0.0105 | 5.0 | 5.5 | 94.48 |
2 | 20 | 0.0060 | 6.5 | 8.0 | 70.03 |
3 | 10 | 0.0060 | 3.5 | 8.0 | 69.30 |
4 | 20 | 0.0060 | 6.5 | 3.0 | 52.02 |
5 | 15 | 0.0105 | 5.0 | 0.5 | 79.38 |
6 | 10 | 0.0060 | 3.5 | 3.0 | 65.40 |
7 | 25 | 0.0105 | 5.0 | 5.5 | 70.70 |
8 | 20 | 0.0060 | 3.5 | 8.0 | 65.01 |
9 | 10 | 0.0150 | 3.5 | 8.0 | 96.58 |
10 | 15 | 0.0105 | 5.0 | 5.5 | 89.80 |
11 | 20 | 0.0150 | 3.5 | 3.0 | 77.81 |
12 | 15 | 0.0105 | 8.0 | 5.5 | 84.16 |
13 | 10 | 0.0150 | 6.5 | 8.0 | 97.85 |
14 | 15 | 0.0105 | 2.0 | 5.5 | 77.51 |
15 | 10 | 0.0060 | 6.5 | 8.0 | 83.95 |
16 | 15 | 0.0015 | 5.0 | 5.5 | 53.03 |
17 | 15 | 0.0105 | 5.0 | 10.5 | 98.48 |
18 | 10 | 0.0150 | 6.5 | 3.0 | 95.10 |
19 | 20 | 0.0150 | 3.5 | 8.0 | 93.20 |
20 | 20 | 0.0060 | 3.5 | 3.0 | 51.02 |
21 | 15 | 0.0105 | 5.0 | 5.5 | 87.85 |
22 | 15 | 0.0105 | 5.0 | 5.5 | 88.46 |
23 | 20 | 0.0150 | 6.5 | 8.0 | 94.30 |
24 | 15 | 0.0195 | 5.0 | 5.5 | 99.50 |
25 | 10 | 0.0060 | 6.5 | 3.0 | 75.90 |
26 | 10 | 0.0150 | 3.5 | 3.0 | 96.77 |
27 | 20 | 0.0150 | 6.5 | 3.0 | 73.50 |
28 | 15 | 0.0105 | 5.0 | 5.5 | 92.10 |
29 | 5 | 0.0105 | 5.0 | 5.5 | 99.87 |
30 | 15 | 0.0105 | 5.0 | 5.5 | 85.90 |
31 | 15 | 0.0105 | 5.0 | 5.5 | 84.40 |
The response may be modeled as:
![]() | (1) |
Source | DF | Adj SS | Adj MS | F-value | P-value |
---|---|---|---|---|---|
Model | 14 | 6098.04 | 435.57 | 34.55 | 0.000 |
Linear | 4 | 5170.51 | 1292.63 | 102.54 | 0.000 |
X1 | 1 | 1097.55 | 1097.55 | 87.06 | 0.000 |
X2 | 1 | 3394.36 | 3394.36 | 269.25 | 0.000 |
X3 | 1 | 69.56 | 69.56 | 5.52 | 0.032 |
X4 | 1 | 609.03 | 609.03 | 48.31 | 0.000 |
Square | 4 | 617.96 | 154.49 | 12.25 | 0.000 |
X12 | 1 | 76.14 | 76.14 | 6.04 | 0.026 |
X22 | 1 | 431.99 | 431.99 | 34.27 | 0.000 |
X32 | 1 | 215.35 | 215.35 | 17.08 | 0.001 |
X42 | 1 | 14.84 | 14.84 | 1.18 | 0.294 |
Interaction | 6 | 309.57 | 51.60 | 4.09 | 0.011 |
X1X2 | 1 | 5.04 | 5.04 | 0.40 | 0.536 |
X1X3 | 1 | 30.09 | 30.09 | 2.39 | 0.142 |
X1X4 | 1 | 180.10 | 180.10 | 14.29 | 0.002 |
X2X3 | 1 | 75.60 | 75.60 | 6.00 | 0.026 |
X2X4 | 1 | 1.69 | 1.69 | 0.13 | 0.719 |
X3X4 | 1 | 17.06 | 17.06 | 1.35 | 0.262 |
Error | 16 | 201.71 | 12.61 | ||
Lack-of-fit | 10 | 129.04 | 12.90 | 1.07 | 0.491 |
Pure error | 6 | 72.66 | 12.11 | ||
Total | 30 | 6299.75 |
The desirability function (DF) was applied to investigate the optimal conditions based on Derringer's desirability function.30,31
The morphology and size of Zn(OH)2-NPs were studied by scanning electron microscopy (Fig. 2b). It can be observed that the average size of nanoparticles is about 60 nm.
The characteristic functional groups of Zn(OH)2-NPs, Zn(OH)2-NP-AC and AC were investigated using the FTIR spectra (Fig. 2c). A peak appearing around 500 cm−1 corresponds to the stretching mode of Zn–O bonds in the spectra of Zn(OH)2-NPs and Zn(OH)2-NP-AC. The broad band at around 3426 cm−1 was assigned to the symmetric and asymmetric O–H stretching vibrations of water present in Zn(OH)2-NPs and/or AC functional groups. The peak appearing at 1500 cm−1 is due to H–O–H bending. In other words, the observed peaks in the range of 1500–3500 cm−1 are probably attributed to either the absorbed water molecules in the KBr matrix, the prepared nanostructures, or their probable interaction. The observed peak around 1100 cm−1 is due to the single C–C bond stretching mode of acetate solution used.
Diffuse reflectance spectrum (DRS) of Zn(OH)2-NPs known as a direct band-gap material was taken over the energy range from 2 eV to 4 eV (Fig. 2d). The energy band gap was determined using a Tauc plot, which is the plot of (αhν)2 versus hν (inset of Fig. 2d). The band-gap value at 25 °C was determined to be 3.34 eV by extrapolating the straight-line portion of (αhν)2 to cross the hν axis, as shown in the inset of Fig. 2d. This value is higher than that of bulk Zn(OH)2 which is 3.2 eV, indicating nanoparticle formation based on the fact that the band gap of each material becomes larger as its dimension decreases.
R%MG = −3.7 − 0.22X1 + 8068X2 + 17.94X3 − 1.82X4 − 0.065X12 − 191937X22 − 1.22X32 + 0.268X1X4 − 322X2X3 | (2) |
The coefficients of determination R2 and adjusted R2 were found to be 0.967 and 0.940, respectively. Fig. 3 shows good agreement between the experimental and calculated removal percentage values.
The profile for predicted values and desirability option (not shown) was used to optimize the adsorption process. Profiling the desirability of response involves specifying the desirability function (DF) for R%MG as a dependent variable by assigning the predicted R%MG values. The scale in the range of 0.0 (undesirable) to 1.0 (very desirable) is used to obtain a comprehensive function that should be maximized according to the efficient selection and optimization of designed variables. The CCD design matrix results (Table 1) show the maximum (99.87%) and minimum (51.02%) of R%MG. Regarding this range, the optimum condition, at which the maximum response is achieved, was obtained. From these calculations and desirability score of 1.0, a maximum recovery of 99.33% was obtained at optimum conditions. The optimal values for the factors MG concentration, adsorbent dosage, pH and sonication time were found to be 20 mg L−1, 0.019 g, 4.5 and 8.6 min, respectively. At this condition, R%MG was predicted to be 99.32% with desirability of 0.99. The validity of the predicted response at optimum conditions was checked by performing three experiments under similar conditions. On average, the experimental response was obtained to be 99.15%, which is in excellent agreement with the predicted value.
MG dye adsorption onto AC and Zn(OH)2-NP-AC was studied at optimum values of variables involved. The results are presented in Table 3. The higher dye removal percentage for Zn(OH)2-NP-AC, even at a very low amount of nanoparticle loading (with a weight ratio of 1:
10), may be attributed to its surface area and the number of reactive atoms.7–11
Adsorbent | AC | Zn(OH)2-NP-AC |
---|---|---|
R%MG | 93.35 | 99.06 |
89.75 | 99.20 | |
91.1 | 99.18 | |
Average R%MG | 91.40 | 99.15 |
The contact time necessary to reach equilibrium depends on the initial dye concentration, pH and amount of adsorbent.32 The amount of adsorbed MG at equilibrium (qe (mg g−1)) was calculated as follows:
qe = (C0 − Ce)V/W | (3) |
3D response surfaces of significant interactions are shown in Fig. 4. The variation of MG removal versus sonication time and initial MG concentration is shown in Fig. 4a, where other variables maintained their optimal values. It can be observed for lower concentrations of MG that shorter time is required for adsorption because of the compromised number of adsorbent vacant sites and dye molecules. For further MG removal, more elapsed time is needed. At higher MG concentrations, the change in adsorption time is due to the decrease in diffusion rate and concentration gradient. Fig. 4b shows the extent to which R%MG is affected by the adsorbent mass and pH. It can be observed that R%MG increases by increasing the adsorbent amount, whereas its variation versus pH does not monotonically increase; thus, the apex of negative parabola occurs at pH 4.5 as an optimum value. By varying the pH, the charges of the adsorbent surface and dye molecules are varied and thus the mechanism and strength of the dye adsorption may be affected. At a lower pH, the proton (H+) is abundant in aqueous media and thus the charge of adsorbent surface and MG molecules becomes more positive. This phenomenon causes an electrostatic repulsive force between the adsorbent and adsorbate, which reduces R%MG. At pH far greater than 4.5, the charge of both adsorbent and adsorbate becomes negative, and thus the electrostatic repulsive force plays a role to reduce the dye adsorption.
![]() | ||
Fig. 4 Response surface for the variation of R%MG against (a) MG concentration and contact time and (b) adsorbent mass and pH. |
Moreover, in the proposed process, the MG dye molecules initially must encounter a boundary layer effect and subsequently diffuse into the porous structure of the adsorbent over a longer contact time. The MG molecule can diffuse to most adsorbent pores, which takes a relatively long time. An increase in the adsorption with increasing adsorbent dosage can be attributed to increased adsorbent surface area and the availability of more adsorption sites.33
Various isotherms such as Freundlich, Langmuir, Dubinin–Radushkevich (D–R) and Temkin models may be applied to fit the equilibrium data of the adsorption system. These isotherms are generally used to fit the experimental equilibrium data to investigate the efficiency and cost-effectiveness of the procedure for the removal of high amounts of pollutants using a small amount of adsorbent.
The Langmuir model35–37 describes a monolayer sorption of a target compound on a homogenous surface, whereas the Freundlich model explains a multi-layer adsorption process.38,39 The correlation coefficient corresponding to each isotherm was obtained as a good criterion for the judgment on the applicability of that isotherm to describe the equilibrium data. From the linear plots of the isotherms, all the respective constants, along with their correlation coefficients, were calculated at several adsorbent masses. The full description of adsorption isotherms and the values of corresponding parameters are presented in Table 4.
Isotherm | Equation | Parameters | Adsorbent (g) | |||
---|---|---|---|---|---|---|
0.014 | 0.019 | 0.024 | ||||
Langmuir | Ce/qe = Ce/Qm + 1/KaQm | The slope and intercept of linear plot of Ce/qe versus Ce provide Qm and Ka, respectively | Qm (mg g−1) | 71.429 | 74.627 | 62.5 |
ka (L mg−1) | 2 | 19.706 | 266.667 | |||
RL | 0.032–0.020 | 0.0034–0.002 | 0.00025–0.00015 | |||
R2 | 0.994 | 0.9997 | 1 | |||
Freundlich | ln![]() ![]() ![]() |
The slope and intercept of linear plot of ln![]() ![]() |
1/n | 0.103 | 0.213 | 0.095 |
KF (L mg−1) | 49.545 | 70.637 | 59.156 | |||
R2 | 0.797 | 0.9 | 0.624 | |||
Temkin | qe = Bl![]() ![]() ![]() ![]() |
B1 and KT are calculated from the slope and intercept of linear plot of qe against ln![]() |
Bl | 5.106 | 10.01 | 3.784 |
KT (L mg−1) | 23![]() |
1012.401 | 4![]() ![]() |
|||
R2 | 0.684 | 0.941 | 0.729 | |||
Dubinin and Radushkevich | ln![]() ![]() ![]() |
The slope and intercept of the linear plot of ln![]() |
Qs (mg g−1) | 58.207 | 70.316 | 56.148 |
K(10−9) | 6.00 | 9.00 | 2.00 | |||
R2 | 0.626 | 0.976 | 0.611 |
In the Freundlich isotherm, the constant of Kf provides information on the bonding energy and is known as the adsorption or distribution coefficient, representing the quantity of dye adsorbed onto the adsorbent. 1/n shows the strength of dye adsorption onto the adsorbent (i.e. surface heterogeneity). By increasing the heterogeneous nature of its surface, its value gets closer to zero. The 1/n value of less than 1 indicates normal Langmuir and cooperative adsorption. The value of 0.213 at optimum condition confirms that the adsorption isotherm follows normal Langmuir.
It can be observed in Table 4 that the correlation coefficient (R2) of the Freundlich model (0.8, 0.9 (optimal condition) and 0.62) was obtained to be lower than that of the Langmuir model (0.99, 1 (optimized condition) and 1), which suggests that the removal process well follows the monolayer adsorption. In addition, correlation coefficients (R2) of the Temkin40 and D–R (applied to estimate the porosity, free energy and the characteristics of the adsorbents)41 models are lower than that of the Langmuir. The high correlation coefficient (R2 > 0.99) with maximum monolayer capacity (74.63 mg g−1) from the Langmuir model proves strong applicability of the Langmuir model to explain the equilibrium data for the MG adsorption onto Zn(OH)2-NP-AC (Table 4).
The experimental kinetic data, obtained at various initial MG concentrations (15, 20 and 25 mg L−1) and optimal adsorbent mass (0.019 g), were fitted to different adsorption kinetic models such as pseudo-first-order,42 pseudo-second-order,43 Elovich44 and intraparticle diffusion45 to determine the rate and mechanism of the adsorption process (Table 5). The definition of the parameters presented in Table 5 are presented in the following manner: in the pseudo-first-order42 and pseudo-second-order models, qe (mg g−1) is the amount of MG adsorbed at equilibrium and qt (mg g−1) is the amount of MG adsorbed at time t. k1 (min−1) and k2 (g mg−1 min−1) are the rate constants of pseudo-first-order and pseudo-second-order adsorption, respectively.
Model | Equation | Calculated parameters | ||||
---|---|---|---|---|---|---|
15 | 20 | 25 | ||||
Dye concentration (mg L−1) | ||||||
First-order kinetic | ln(qe − qt) = −k1t + ln(qe) | The slope and intercept of linear plot of ln(qe − qt) versus t provide k1 and qe, respectively | k1 | 2.024 | 2.213 | 2.332 |
qe (calc) | 205.873 | 2026.749 | 5473.940 | |||
R2 | 0.603 | 0.530 | 0.650 | |||
Second-order kinetic | t/qt = t/qe + 1/(k2qe)2 | The slope and intercept of linear plot of t/qt versus t provide qe and k2, respectively | k2 | 0.4885 | 0.080 | 0.027 |
qe (calc) | 39.68 | 53.19 | 64.94 | |||
R2 | 1 | 0.999 | 0.998 | |||
Intraparticle diffusion | qt = Kdifft1/2 + C | The slope and intercept of linear plot of qt versus t1/2 provide Kdiff and C, respectively | Kdiff | 0.710 | 2.826 | 8.287 |
C | 37.49 | 43.82 | 38.25 | |||
R2 | 0.868 | 0.972 | 0.952 | |||
Elovich | qt = 1/β![]() ![]() |
β and α are obtained from the slope and intercept of the plot of qt versus ln(t), respectively | β | 1.500 | 0.398 | 0.131 |
R2 | 0.925 | 0.930 | 0.966 | |||
α | 4.14 × 1024 | 3.48 × 1030 | 1.97 × 1030 | |||
qe(exp) | 39.367 | 52.130 | 60.926 |
The kinetic model for the adsorption of a solute by a solid in an aqueous solution is generally complex. The adsorption rate is strongly influenced by several parameters related to the state of the solid (generally with a highly heterogeneously reactive surface) and to the physicochemical conditions under which the adsorption occurs. The consistency between the experimental and the model-predicted data was investigated by calculating correlation coefficients (R2) and by observing the extent to which the experimental adsorption capacity is close to the theoretical value. The equilibrium adsorption capacity, qe, was observed to increase from 39.68 mg g−1 to 64.98 mg g−1 by increasing the MG concentration from 15 mg L−1 to 25 mg L−1 at 0.0195 g of Zn(OH)2-NP-AC.
High inconsistency between the experimental data and the pseudo-first-order model was observed (Table 5). In pseudo-second-order model, the calculated qe values are very close to that of the experimental data and R2 values were found to be very close to 1.0, indicating that the MG adsorption onto Zn(OH)2-NP-AC obeys the pseudo-second-order kinetic model for the entire sorption period. The rate-constant values obtained from the pseudo-second-order kinetic model were found to decrease from 0.4885 to 0.027 g mg−1 min−1 by increasing the initial MG concentration from 15 to 25 mg L−1 at 0.0195 g of adsorbent.
The linear relation of initial dye concentration with removal rate fails when pore diffusion is the predominant stage and limits the adsorption process. The dye transfer from the bulk to the solid phase may occur through an intraparticle diffusion/transport process known as the rate-limiting step.46 In the intraparticle diffusion model, the value of intercept C (Table 5) provides information about the thickness of the boundary layer and the external mass transfer resistance. The constant C was found to be 37.49, 43.82 and 38.25 at MG concentrations of 15, 20 and 25 mg L−1, respectively, whereas the Zn(OH)2-NP-AC mass was set to be 0.0195 g. This variation is attributed to the increase in the boundary layer thickness, which decreases the chance of external mass transfer and subsequently increases the amount of internal mass transfer. In addition to the applicability of the pseudo-second-order kinetic model, the intraparticle diffusion model was also found to be applicable to explain the experimental data (Table 5) with high R2 value (0.97 at an optimal adsorbent mass). This may confirm that the rate-limiting step is the intraparticle diffusion process. The intraparticle diffusion rate-constant (Kdiff) values were obtained to be 0.710, 2.826 and 8.287 mg g−1 min−1/2 at 0.019 g of adsorbent with good positive correlation to the initial dye concentration. This linear relationship shows a high contribution of intraparticle diffusion on the adsorption process.
Adsorbent | Adsorption capacity (mg g−1) | Contact time (min) | References |
---|---|---|---|
AC | 4.34 | 89 | 26 |
Activated carbon | 75.08 | 1440 | 47 |
Activated slag | 74.16 | 1440 | 47 |
Cyclodextrin-based material | 91.9 | 120 | 48 |
Bentonite clay | 7.72 | 10 | 49 |
Commercial activated carbon | 8.27 | 15 | 50 |
Laboratory grade activated carbon | 42.18 | 15 | 50 |
Arundo donax-root carbon | 8.69 | 180 | 51 |
ZnO-NP-AC | 76.92 | 15 | 52 |
ZnO-nanorod-AC | 59.17 | 4 | 53 |
CoFe2O4-AC | 20 | 89.3 | 54 |
Pisum sativum | 35 | 6.17 | 55 |
Zn(OH)2-NP-AC | 74.63 | 8.6 | This work |
This journal is © The Royal Society of Chemistry 2015 |