Water defluoridation by aluminium oxide--manganese ...

This article was downloaded by: [Addis Ababa University] On: 22 April 2014, At: 03:37 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Environmental Technology Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/tent20

Water defluoridation by aluminium oxide–manganese oxide composite material a

a

a

Sheta Alemu , Eyobel Mulugeta , Feleke Zewge & Bhagwan Singh Chandravanshi

a

a

Department of Chemistry, Faculty of Science, Addis Ababa University, PO Box 1176, Addis Ababa, Ethiopia Published online: 28 Feb 2014.

To cite this article: Sheta Alemu, Eyobel Mulugeta, Feleke Zewge & Bhagwan Singh Chandravanshi (2014) Water defluoridation by aluminium oxide–manganese oxide composite material, Environmental Technology, 35:15, 1893-1903, DOI: 10.1080/09593330.2014.885584 To link to this article: http://dx.doi.org/10.1080/09593330.2014.885584

PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions

Environmental Technology, 2014 Vol. 35, No. 15, 1893–1903, http://dx.doi.org/10.1080/09593330.2014.885584

Water defluoridation by aluminium oxide–manganese oxide composite material Sheta Alemu, Eyobel Mulugeta, Feleke Zewge∗ and Bhagwan Singh Chandravanshi Department of Chemistry, Faculty of Science, Addis Ababa University, PO Box 1176, Addis Ababa, Ethiopia

Downloaded by [Addis Ababa University] at 03:37 22 April 2014

(Received 11 May 2013; accepted 13 January 2014 ) In this study, aluminium oxide–manganese oxide (AOMO) composite material was synthesized, characterized, and tested for fluoride removal in batch experiments. AOMO was prepared from manganese(II) chloride and aluminium hydroxide. The surface area of AOMO was found to be 30.7 m2 /g and its specific density was determined as 2.78 g/cm3 . Detailed investigation of the adsorbent by inductively coupled plasma-optical emission spectrometry, inductively coupled plasmamass spectrometry, and ion chromatography (for sulphate only) showed that it is composed of Al, Mn, SO4 , and Na as major components and Fe, Si, Ca, and Mg as minor components. Thermogravimetric analysis was used to study the thermal behaviour of AOMO. X-ray diffraction analysis showed that the adsorbent is poorly crystalline. The point of zero charge was determined as 9.54. Batch experiments (by varying the proportion of MnO, adsorbent dose, contact time, initial F− concentration, and raw water pH) showed that fluoride removal efficiency of AOMO varied significantly with percentage of MnO with an optimum value of about 11% of manganese oxide in the adsorbent. The optimum dose of the adsorbent was 4 g/L which corresponds to the equilibrium adsorption capacity of 4.8 mg F− /g. Both the removal efficiency and adsorption capacity showed an increasing trend with an increase in initial fluoride concentration of the water. The pH for optimum fluoride removal was found to be in the range between 5 and 7. The adsorption data were analysed using the Freundlich, Langmuir, and Dubinin–Radushkevich models. The minimum adsorption capacity obtained from the non-linear Freundlich isotherm model was 4.94 mg F− /g and the maximum capacity from the Langmuir isotherm method was 19.2 mg F− /g. The experimental data of fluoride adsorption on AOMO fitted well to the Freundlich isotherm model. Kinetic studies showed that the adsorption is well described by a non-linear pseudo-second-order reaction model with an average rate constant of 3.1 × 10−2 g/min mg. It is concluded that AOMO is a highly promising adsorbent for the removal of excess fluoride from drinking water. Keywords: fluoride; AOMO; defluoridation; manganese oxide; aluminium oxide

1. Introduction Natural and anthropogenic activities can enhance the levels of fluoride in the environment. Volcanoes represent the main natural persistent source of fluoride, whereas anthropogenic activities involve extraction and processing of minerals, the phosphate rock industry and waste, and application of phosphate fertilizers.[1] Fluoride has certain physiological properties of great interest in relation with the human health and well-being. [2,3] Although a low daily dose of fluoride is well known to be useful for preventing the formation of dental caries, exposure to excess fluoride can have harmful effects on human health and could result in dental fluorosis or skeletal fluorosis, depending on the amounts of fluoride intake.[4] Mottling of teeth is one of the earliest and most easily recognized symptoms of dental fluorosis. In severe fluorosis some victims can experience deformation of bones (skeletal fluorosis) and eventually becomes crippling. Prevalence of dental and skeletal fluorosis has been reported in several parts of the world including Ethiopia, where fluoride concentration in drinking water exceeded the guideline ∗ Corresponding

author. Email address: [email protected]

© 2014 Taylor & Francis

level.[5–8] However, up to 33 mg/L of F− concentrations were reported in water samples from boreholes in Ethiopia. Such high levels are found in the Rift Valley region, which is characterized by a relatively high volcanic activity in the country. Possible control options to protect fluorosis may include provision of alternative source of water, blending with low fluoride-containing water, harvesting rainwater, and provision of bottled water, at least for young persons. When none of the above options is feasible or if the only solution would take a long time for planning and implementation, defluoridation of drinking water has to be practiced. Several attempts have been made to reduce the fluoride to an acceptable level in drinking water over the years, using a wide variety of materials. At present, there are few treatment methods that are used for controlling excessive levels of fluoride in drinking water. Based on the mechanism of fluoride removal, the methods can be categorized into chemical precipitation by lime and alum;[9] adsorption by activated alumina, clay materials, and industrial waste residues;[10–16] ion exchange by membranes, synthetic


1894

S. Alemu et al.

resins, and bone char;[17–22] and by membrane technologies such as reverse osmosis and electro-dialysis.[23,24] Among these methods, adsorption is still one of the most extensively used methods for the defluoridation of water due to its low cost and viability.[25,26] The methods used by industrialized countries, such as reverse osmosis, electro-dialysis, and ion-exchange, require more technical support for operation and maintenance and the capital investment cost is very high.[9] Despite such efforts, rural communities in the several regions of developing countries are consuming water with fluoride content exceeding the recommended level. In previous studies, the fluoride removal capability of aluminium hydroxide has been demonstrated.[27] Results showed that the material has an adsorption capacity of about 23 mg/g. However, the applicability of the material is limited under continuous operations due to the relatively soft nature of the material. In recent years, studies have shown that oxides of manganese have a good ability to remove anions from aquatic environments.[28,29] Other studies showed that the defluoridation performance of activated alumina is significantly improved when coated with manganese oxide.[30–32] In this study, aluminium oxide–manganese oxide (AOMO) composite material has been synthesized, characterized, and its defluoridation potential was assessed in batch sorption experiments. The adsorption isotherm models and its kinetics have been also investigated in the batch mode. 2.

Materials and methods

2.1. Synthesis of the adsorbent Hydrated aluminium sulphate (Al2 (SO4 )3 · 14H2 O), which is locally available in Ethiopia, was used in the preparation of aluminium hydroxide, AOH (which is used as a base material for the synthesis of AOMO). In a typical procedure, 100 g of aluminium sulphate was dissolved in 500 mL of deionized water, and the resulting lower pH (about 2.7) was adjusted to pH 7.0 with 2 M NaOH. The precipitate was separated and dried in an oven for 12 h at 100◦ C. Then, the same material was placed into a furnace (Calbolite, ELF Model, Waglech International Ltd., UK) at 300◦ C for 1 h to produce aluminium hydroxide (AOH).[27] Then, the adsorbent (AOMO) was prepared in two steps.[31,32] In the first step, a 50 mL solution mixture (4.58 g of MnCl2 · H2 O and 0.5 mL of 10 M NaOH) was poured over 50 g of aluminium hydroxide in a heatresistance dish, and heated to 150◦ C for about 5 h in an oven. In the second step, the same mixture was again heated at 500◦ C for 3 h in a furnace. 2.2. Fluoride analysis A fluoride stock solution (1000 mg/L) was prepared from 99.0% NaF (BDH Chemicals, England) in deionized water.

Standards and samples were prepared by appropriate dilution of the stock solution. The fluoride concentration was measured with A pH/ISE (ion selective electrode) meter (Orion model, EA 940 Expandable Ion Analyzer, USA) equipped with combination fluoride selective electrode (Orion Model 96-09, USA). Analyses were performed on equal-volume mixtures of sample and total ionic strength adjustment buffer, the latter being required to suppress interferences. The pH was measured with pH/ION meter (WTW Inolab pH/ION Level 2, Germany) using an unfilled pH glass electrode. 2.3. Characterization of adsorbent The samples (0.1 g) were digested in a microwave digester in a 3:1 mixture (4 mL) of 30% HCl and 65% HNO3 for 80 min and then diluted to 100 mL using deionized water. The elemental composition (Li, B, Na, Mg, Al, Si, P, Cl, K, Ca, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Sr, Mo, Cd, Sn, Sb, Ba, La, W, Au, Hg, Ti, Pb, Th, U, Br, Bi, Cs, Nb) was determined by inductively coupled plasma-optical emission spectrometry (ICP-OES) (SPECTRO CIROSCCD , Germany) inductively coupled plasma-mass spectrometry (ICP-MS) (Agilent 7500CX, USA), and sulphate was determined by ion chromatography (Metrohm 761, Switzerland). The absolute density was determined using the Pycnometer (MicroMeritics AccuPyc 1330, USA). The surface area (SA) of the samples was determined by the N2 adsorption method (Thermofinnigan Sorptomatic 1900, Germany). The sample was first outgassed at 110◦ C and then allowed to cool to room temperature. Thermogravimetric analysis (TGA) was used to measure the weight loss of the samples after placing them in a ceramic crucible and heated at a rate of 10◦ /min from room temperature to 1000◦ C. The instrument used was Mettler Toledo AG-TGA/SDTA851e, Switzerland. The X-ray diffraction (XRD) pattern was recorded by an X-ray diffractometer (X’pert PRO, PANalytica, The Netherlands) using Cu Kα radiation (λ = 0.1542 nm) with a 0.05◦ /min step scanned from 5◦ to 80◦ in 2θ angle. Scanning electron microscopy (SEM) analysis was carried out using a NOVA NANOSEM (FEI Company, USA) equipped with GAD detector, with an accelerating voltage of 5 kV. The point of zero charge (PZC) was measured by the potentiometric mass titrations technique.[33] PZC was identified as the common intersection point of the potentiometric curve of the blank solution with the corresponding curves of the impregnating suspensions containing 0.5, 1.0, and 1.5 g of AOMO in electrolytic solution (0.02 N NaNO3 in 50 mL of deionized water). The experiment was performed under an N2 atmosphere and the aqueous suspensions were equilibrated for 1 h to reach an equilibrium pH value. Small amount of 1 M NaOH was added to make the pH around 10 and recorded as initial pH after 15–20 min. Then the solid suspensions were titrated by 0.1 N HNO3 , using 665 Dosimat (Metrohm, Switzerland). The pH of each suspension is then measured with

Environmental Technology


1 min time interval using a digital pH meter standardized by buffers (WTW Inolab pH/ION Level 2, Germany). 2.4. Batch adsorption studies Batch experiments were conducted in a 500 mL Erlenmeyer flask under continuous mixing condition with magnetic stirrers at room temperature (23 ± 2◦ C). A sample was taken as required, filtered, and analysed immediately for its fluoride content. The defluoridation capacity and percentage of adsorption at a given time under specified conditions were determined based upon the measurement of the liquid phase concentration. To assess the effect of varying proportions of manganese oxide on defluoridation performance of AOMO, the amount of MnCl2 · H2 O in the solution was varied (8.8, 13.7, 19.1, and 25.2 g) in order to obtain 11%, 16%, 21%, and 26% manganese oxide in AOMO, respectively. Four grams per litre of each adsorbent with 0%, 6%, 11%, 16%, 21%, and 26% of manganese oxide was mixed with 20 mg/L fluoride solution at a contact time of 3 h. To investigate the effect of dose and contact time, experiments were conducted by varying adsorbent doses within the range between 0.8 and 4.8 g/L using an initial fluoride concentration of 20 mg/L and a contact time of 3 h. The effect of initial fluoride concentration and contact time were studied by altering the initial fluoride concentrations as 5, 10, 15, 20, and 25 mg/L using the constant adsorbent dose of 4 g/L. The effect of raw water pH was investigated with an initial fluoride concentration of 20 mg/L and the adsorbent dose of 4 g/L by varying the initial solution pH from 3 to 10, using 0.1 M HCl or 0.1 M NaOH. Adsorption isotherm experiments were conducted using an adsorbent dose of 4 g/L and varying initial fluoride concentrations within the range between 5 and 70 mg/L at constant pH of 7. Kinetic study of the adsorption data is based on pseudo-first-order and pseudo-second-order reaction rates. Adsorption kinetics was determined using constant surface loadings of 1, 2, and 4 g/L corresponding to the initial fluoride concentrations of 10, 20, and 40 mg/L, respectively. Linear and non-linear regression methods were used to compare adsorption isotherm and kinetic parameters and applicability of the models. Three isotherm models (Freundlich, Langmuir, and Dubinin–Radushkevich (D–R)) and two kinetics (pseudo-first-order and pseudo-second-order) models were used. Graph Pad Prism versions 6.0 software was used for the non-linear regression analysis. Table 1.

3. Results and discussion 3.1. Characterization of adsorbent The elemental composition of AOMO in comparison with aluminium hydroxide is summarized in Table 1. The major components that make up more than 97% of the solid phase (AOMO) are aluminium, manganese, sodium, sulphate, and the minor components are iron, silicon, potassium, calcium, and magnesium. (Assuming Al to be present as Al(OH)3 ). The remainder is primarily composed of Fe (13.9 mg/g). All other elements were present in concentrations below 3.0 mg/g. Therefore, the presence of sulphate, manganese, and small amount of iron in AOMO would contribute to the higher fluoride uptake capacity. The presence of sulphate that is associated with Al is responsible for the acidity of the adsorbent and hence high fluoride uptake by AOMO, whereas sulphate content associated with Na might be further reduced if the washing efficiency improved. The AOMO material appears dark brownish (brown–black) in colour since it consists of Mn and Fe as manganese and iron oxides. Figure 1 shows the TGA curve of AOMO. The first step’s ranges from 70◦ C to 350◦ C with a weight loss of 2.8% corresponding with the evolution of all the water (70–120◦ C) and the formation of manganese dioxide and aluminium oxyhydroxide compounds (120–350◦ C).[34] The second step ranges from 640◦ C to 980◦ C with a weight loss of 10.8% corresponding with the reduction of manganese dioxide to manganese trioxide and complete decomposition of aluminium oxyhydroxide to aluminium oxide (640–760◦ C); and also represents the conversion of Mn2 O3 to Mn3 O4 and the formation of MnAl2 O4 as the result of solid–solid interaction between the corresponding oxides.[35] The TGA result was supported later by the XRD investigation of the sample. It can be seen from the XRD pattern (Figure 2(a)) that poorly crystalline MnO2 and Al2 O3 phases were detected. This might be attributed to the presence of aluminium oxide that leads to an increase in the degree of dispersion of MnO2 on its surface thus hindering their grain growth.[35] The SEM image is shown in Figure 2(b), and AOMO had a significantly rough surface porous. Thus, presence of MnO2 increases the number of active sites and plays an important role in the fluoride adsorption process.[36] The measurement of PZC showed that the surface of AOMO is positively charged when solution pH is below its PZC (9.54), facilitating fluoride adsorption through the electrostatic attraction between fluoride and adsorbent.

Elemental composition, SA, and density of AOMO and AOH.

Adsorbent

Concentration

AOMO

mg/g m mol/g mg/g m mol/g

AOH

1895

Al

Mn

Na

SO4

255 9.44 288 10.7

122 2.22 0.39 0.01

102 4.43 31 1.35

85.8 0.89 155 1.61

SA (m2 /g)

Density (g/cm3 )

30.7

2.78

37.7

2.39

1896

S. Alemu et al. 0.04

102 100

TG DTG

0.02

Weight loss (%)

96 94

0.00

92 90

-0.02

88 86

st 1 Derivative of TG

98

-0.04

84 82 -0.06 0

200

400

600

800

1000


Temperature (ºC)

Figure 1. Thermogravimetric (TG) curve (dotted line) and derivative of thermogravimetric (DTG) curve (solid line) of AOMO. (a) 400 300 300 250 200 150 100 50 0

0

10

20

30

40 50 60 Position {2 Theta]

70

80

90

(b)

Figure 2. (a) X-ray diffractogram of AOMO and (b) SEM images of AOMO.

3.2.

Effect of process parameters on fluoride removal efficiency by AOMO

3.2.1. Effect of the proportion of manganese oxide As it can be seen from Figure 3, fluoride removal efficiency of AOH-300 (aluminium hydroxide treated at 300◦ C) is much larger than that of AOH-500. This is due to the fact that AOH-500 has lost a considerable amount of hydroxyl groups compared with AOH-300, which accounts for the reduction in fluoride removal efficiency. It can also be

Figure 3. Effect of thermal treatment and percentage of manganese oxide on fluoride removal efficiency of AOMO (adsorbent dose = 4 g/L, initial fluoride concentration = 20 mg/L, contact time = 180 min, and pH = 7.0 ± 0.20).

seen that fluoride removal efficiency is higher with manganese oxide proportion ranging from 11% to 16% and then decreased for 21% and 26%. An increase in fluoride removal efficiency with percentage of manganese oxide in the adsorbent is possibly due to the formation of porous manganese oxide on the outside and inner surfaces of aluminium oxide. As more and more manganese oxide is formed especially on the inner surfaces of the base material, SA available per unit mass of the adsorbent for the adsorption may decrease and may result in low fluoride removal efficiency. An adsorbent with 11% manganese oxide showed a relatively greater adsorption efficiency, thus used for further study as AOMO.

3.2.2. Effect of adsorbent dose and contact time The effect of adsorbent dose and contact time on the fluoride removal efficiency is shown in Figure 4. It can be seen that the residual fluoride concentration decreases as the adsorbent dose increases. The rate of removal of fluoride is fast during the first 5 min. After 30 min, the rate of removal of fluoride decreases and reaches equilibrium within 120 min. The time to reach equilibrium appears to be independent of adsorbent dose under the experimental conditions used for this study. However, after 30 min the rate of change in fluoride concentration was smaller for higher adsorbent doses (≥3.2 g/L). The fluoride removal efficiency and capacity as a function of adsorbent dose are shown in Figure 5. The increase in fluoride removal efficiency was possibly due to the increase in availability of fluoride-binding sites resulting from an increase in adsorbent dosage.[16] When the adsorbent dose was increased, beyond 4 g/L, there was no significant change in the percentage of fluoride removed. On the other hand, the adsorption capacity decreases with increasing dose. To maintain maximum capacity and high removal efficiency, the surface loading (i.e. the mass ratio of fluoride to adsorbent dose) should be lower than



Figure 4. Time concentration profile at different doses of AOMO (initial concentration = 20 mg/L and initial solution pH = 7.0 ± 0.2).

1897

Figure 6. Effect of initial solution pH on fluoride removal efficiency (initial F− concentration = 20 mg/L, adsorbent dose = 4 g/L, and contact time = 120 min).

the Kd values at a given pH should not change with the adsorbent dose.

Figure 5. Adsorption capacity and % of fluoride removal as a function of adsorbent dose (initial F− concentration = 20 mg/L, contact time = 120 min, and pH = 7.0 ± 0.2).

the optimum value. The surface loading for optimum fluoride removal obtained from Figure 5 is less or equal to 5 mg/g. A dose of 4 g/L, corresponding to the capacity of 4.8 mg F− /g of the adsorbent, was considered for further adsorption studies. A distribution coefficient Kd , for fluoride adsorbed on the adsorbent at pH 7 was calculated[37] Cs (L/g), Kd = Cw

(1)

where, Cs is the concentration of fluoride on the adsorbent (mg/g) and Cw is the concentration of fluoride in water (mg/L). The concentration of fluoride in the solid phase was calculated from the measurement of fluoride in the water before and after adsorption.[37] The value of Kd as a function of adsorbent dose (data not shown) indicates that the Kd value increases linearly (R2 = 0.975) with an increase in the adsorbent dose at a constant pH which may suggest the heterogeneous nature of the adsorbent surface. If the surface is homogeneous,

3.2.3. Effect of initial fluoride concentration The effect of initial fluoride concentration on the adsorption of fluoride by the adsorbent at equilibrium was studied at a constant adsorbent dose of 4 g/L. There is no significant difference on the fluoride removal percentage at equilibrium with an increase in initial fluoride concentrations (data not shown), indicating that the material can be used in wide fluoride concentration ranges. However, residual concentration at equilibrium increases with an increase in initial fluoride concentration. On the other hand, the adsorption capacity increases with an increase in the initial concentration. This may be due to the availability of a higher number of fluoride ions per unit mass of the adsorbents and/or the utilization of less accessible or energetically less-active sites because of increasing diffusivity of fluoride ions when initial concentration increases.[25]

3.2.4. Effect of raw water pH The pH of the solution has been identified as the most important variable governing adsorption.[38] The experimental results of the effect of raw water pH on the adsorption of fluoride are shown in Figure 6. Strong dependence of removal efficiency on pH is observed within a pH range studied. Surface charge of an oxide mineral in aqueous systems will change with changing pH and would be hydroxylated to develop a surface condition in which there is an uneven charge distribution over the surface.[39,40] Thus, significant and rapid removal of fluoride in the acidic pH range and decreased removal at high pH can be explained due to the ion-exchange adsorption mechanism. Protonation of the oxide surface followed adsorption of fluoride through ligand exchange can explain the overall adsorption process.

1898

S. Alemu et al. Table 2.

Linear and non-linear forms of isotherm equations and parameters.

Isotherms Langmuir

Linear Ce 1 = + qe qm Parameters qm (mg/g) 18.62

Freundlich


D–R

qe = R2 0.9712

b (L/mg) 0.3834

1 log(Ce ) n

qm (mg/g) 19.19

b (L/mg) 0.3300

R2 0.9772

N 1.946

R2 0.9895

1/n

R2 0.9529

n 1.668

qm bCe 1 + bCe

qe = KF Ce

ln qe = ln qs − βε 2 Parameters qs (mg/g) 99.87

Table 3.

Ce qm

log(qe ) = log(KF ) + Parameters KF (mg/g) 4.481

Non-linear

KF (mg/g) 4.942

qe = (qs ) exp (−βε 2 ) R2 0.9645

E (kJ/mol) 9.71

qs (mg/g) 16.21

E (kJ/mol) 27.08

R2 0.9528

Linear and non-linear forms of kinetic model equations and parameters.

Kinetic Pseudo-first-order Initial F− and adsorbent dose 10 mg/L F− and 1 g/L dose 20 mg/L F− and 2 g/L dose 40 mg/L F− and 4 g/L dose Pseudo-second-order Initial F− and adsorbent dose 10 mg/L F− and 1 g/L dose 20 mg/L F− and 2 g/L dose 40 mg/L F− and 4 g/L dose

Linear

Non-linear

log(qe − qt ) = log qe −

K1 t 2.303

Parameters qe (mg/g) K1 1.893 7.28 × 10−3 1.639 9.92 × 10−3 1.334 7.73 × 10−3 1 1 t = + t qt qe K2 qe2 Parameters qe (mg/g) K2 8.73 2.20 × 10−2 9.29 3.67 × 10−2 9.19 4.54 × 10−2

The concentration of aluminium was in the range of 0.013– 0.06 mg/L in the pH range between 6.5 and 8.0, which is below the WHO recommended value for drinking water. Figure 6 shows that initially the adsorption of fluoride increased from pH 3 to 5 reached a maximum of about 96.4% at pH 5. Then a very slight change in fluoride removal was observed up to pH around 7.0. The high removal efficiency in the acidic pH range is due to the existence of positive sites and neutral sites on the surface of the adsorbent that facilitates for more fluoride ions to bind on the surface resulting in high fluoride removal efficiency. After pH 8, fluoride adsorption decreased sharply and as low as 61.1% was removed at pH 10. The low adsorption efficiency of the media at high pH value is attributed to the strong competition of hydroxide ions on the adsorbent surface for the adsorption site as well as the repulsion of fluoride ions by the negatively charged surface of the adsorbent.[16] At lower pH, the adsorption efficiency is less, which is possibly due to the formation of weakly ionized hydrofluoric acid.

qt = qe (1 − exp−K1 t ) R2 0.8224 0.8952 0.7895

qe (mg/g) K1 8.442 0.1119 9.038 0.1674 9.383 0.1939 K2 qe2 t qt = 1 + K2 qe t

R2 0.9812 0.9838 0.9858

R2 0.9999 0.9998 0.9998

qe (mg/g) 8.726 9.285 9.615

R2 1.0000 1.0000 1.0000

K2 0.0220 0.0324 0.0374

3.3. Adsorption isotherms and kinetics The determined coefficients for isotherm and kinetic models are summarized in Tables 2 and 3, respectively. Figures 7 and 8 were plotted using the experimental and predicted value by linear and non-linear regression methods for isotherm and kinetics models, respectively. The error analysis was also performed using six different error functions (absolute sum of squares (ASS), sum of the squares for the residuals (Sy.x ), chi-square (χ 2 ), average percentage errors (APE), hybrid fractional error function (HYBRID), and Marquardt’s percent standard deviation (MPSD)) to evaluate the applicability of each isotherm model equation to the experimental data (Table 4). From Table 2, it was observed that the correlation coefficient value for the linear form of the Langmuir isotherm is higher (R2 = 0.9712) than Freundlich and D–R isotherms. However, the non-linear isotherm correlation coefficient for Freundlich (R2 = 0.9895) is higher than the linear as well as the non-linear forms of Langmuir and D–R



1899

Figure 7. Linear and non-linear forms of isotherm models for the adsorption of fluoride by AOMO (a) linearized Langmuir, (b) linearized Freundlich, (c) linearized D–R and (d) non-linear isotherms (adsorbent dose = 4 g/L, contact time = 4 h, and pH = 7.0 ± 0.2).

isotherms. Therefore, the non-linear model could be the best representative of the experimental data as compared with linear models. It is clear that transformations of the non-linear Langmuir isotherm equation to linear forms implicitly alter the error structure and may also violate the error variance and normality assumptions of the standard least-squares method.[41] The error analysis performed for non-linear isotherms also indicates that the Freundlich isotherm was the one with the lowest value (Table 4). Thus, from the values of correlation coefficient and error analysis, the non-linear Freundlich isotherm model was found preferable to explain the experimental results of fluoride adsorption onto AOMO. The applicability of the Freundlich isotherm model suggested that intercalation of F− into the interlayer and adsorption of F− on the external surface should be involved in the process of fluoride removal by AOMO.[4] This study confirms that it is highly irrelevant to use the linear method to obtain the parameters in the Lagergren pseudo-first-order kinetic expression. By the linear method pseudo-second-order kinetics very well represent the kinetic uptake of fluoride onto AOMO. The nonlinear method was found to be a better method than the linear method for predicting the optimum kinetics and the parameters involved in them. By the non-linear method

both pseudo-first-order and pseudo-second-order kinetics very well represent the kinetics of fluoride onto AOMO. However, from error analysis (Table 4) and correlation coefficient values, it is evident that the second-order kinetics is the best representative for the adsorption of fluoride onto AOMO. The rate constants for the three initial concentrations were very close and therefore, the three constants were averaged to obtain a single rate constant of 3.1 × 10−2 g/min mg.

3.4. Intraparticle diffusion Even though adsorption is a surface phenomenon, the adsorbate may also diffuse into the interior pores of the adsorbent, which may influence the rate of reaction. In order to check the contribution of intraparticle diffusion, an intraparticle diffusion model [42] was used. Three different regions can be identified in Figure 9. The first, sharp portion of the curve corresponds to the external surface adsorption stage or instantaneous adsorption stage. The second portion of the curve indicates the intraparticle diffusion, which is a predominant rate-controlling process. Under this condition, the diffusive transport of fluoride ions occurs through the internal pores of the adsorbent. The third


1900

S. Alemu et al.

Figure 8. (a) Linear and (c) non-linear pseudo-first-order and (b) linear and (d) non-linear pseudo-second-order kinetics for the adsorption of fluoride on AOMO with initial fluoride concentrations of 10 mg/L, 20 mg/L, and 40 mg/L, to adsorbent doses of 1.0, 2.0, and 4.0 g/L, respectively (pH = 7.0 ± 0.2 and contact time = 10 h).

due to the heterogeneous nature of the adsorbent. Therefore, the adsorption of fluoride onto AOMO is influenced by both surface reactions as well as intraparticle diffusion effects.

Figure 9. Adsorption capacity (mg/g) as a function of square root of time (min1/2 ).

region becomes dominant if the bulk fluoride concentration is low. Linear portion of the curve did not pass through the origin (not shown), which indicates the complex nature of the adsorption process. This unusual behaviour is possibly

3.5. Comparison of defluoridation capacities Comparison of defluoridation capacity of different sorbents is essential to evaluate their relative performance. Comparison of adsorption capacity was made for selected sorbents. The results in Table 5 clearly indicate that AOMO has a much higher defluoridation capacity than other adsorbents except AOH. The use of AOMO as a defluoridation media could have advantage over AOH in that its strength is much better for application in the continuous packed bed column. Comparison of defluoridation capacities and kinetics of adsorption shows AOMO is a promising sorbent for removal of excess fluoride from drinking water.

Environmental Technology Table 4.

1901

Comparison of linear and non-linear models using error functions. Isotherm models

Error functions ASS =

Regression methods

n

(qe,cal − qe,meas )2

Kinetic models

Langmuir

Freundlich

D–R

Pseudo-first order

Pseudo-second order

Linear

0.0074

0.0521

0.2049

1.431

0.6289

Non-linear

4.569

2.110

9.464

1.353

0

Linear

0.0304

0.0807

0.1600

0.3441

0.2241

Non-linear

0.7557

0.5136

1.088

0.3356

0

Linear

0.0344

0.4364

−0.023

−0.2067

0.1143

Non-linear

0.8460

0.9784

1.608

0.3207

0.0011

i=1

Sy.x =

SS df

2 n (q e,cal − qe,meas ) = qe,meas i=1 n q 100 e,meas − qe,cal APE = n i=1 qe,meas


χ2

n (q 100 e,meas − qe,cal ) HYBRID = n − p i=1 qe,meas

MPSD = 100

1 n−p

n i=1

qe,meas − qe,cal qe,meas

2

Linear

10.79

29.09

1.2

Non-linear

10.07

12.67

17.1

Linear

−3.777

Non-linear Linear Non-linear

6.653 259.2 2278

−25.60

4.731

−0.019

−9.432

9.7982

102.9

4.284

44.79

65.85

66.62 2.046 232.9

60.32

16.73

3.9777 0.1423 −2.404 −0.174 14.06 0.503

Note: The bold values indicate the lowest value within three isotherm and two kinetic models, respectively. Table 5.

Defluoridation capacities of aluminium oxide-based and other adsorbents.

Adsorbents Acidic alumina Activated alumina (Grade: OA-25) Activated alumina (Grade: AD101-F) Modified imobillized activated alumina AOH Manganese dioxide coated activated alumina Manganese-oxide-coated alumina Magnesia-amended activated alumina Manganese-oxide-modified activated alumina Alum sludge Hydroxyapatite Nanosized fluorapatite Nano alumina AOMO

4. Conclusions In this study, the capability of AOMO for fluoride removal was demonstrated. The fluoride uptake varied depending on the percentage of manganese oxide. Under optimum manganese oxide to aluminium hydroxide proportion, factors such as the dose of the adsorbent, initial concentration of, and raw water pH influenced fluoride uptake. The adsorbent was found to be very efficient in removing fluoride in a pH range of 5–7. Although all the studied isotherm and kinetics models showed fairly good fit to the experimental data, the Freundlich isotherm model and Lagergren second-order kinetic model were found to be the best ones to explain fluoride adsorption onto AOMO. AOMO has a much higher

Adsorption capacity(mg/g)

References

8.4 1.78 0.415 0.76 23.70