The removal of copper in water using manganese activated saturated and unsaturated sand filters

Chemical Engineering Journal 209 (2012) 334–344

Contents lists available at SciVerse ScienceDirect

Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej

The removal of copper in water using manganese activated saturated and unsaturated sand filters Chedly Tizaoui ⇑, Sugihhartati Dj Rachmawati, Nidal Hilal Centre for Water Advanced Technologies and Environmental Research (CWATER), College of Engineering, Swansea University, SA2 8PP, UK

h i g h l i g h t s

g r a p h i c a l a b s t r a c t

" Copper is widely distributed in the

1

q/qsat

i n f o

Article history: Received 22 May 2012 Received in revised form 2 August 2012 Accepted 3 August 2012 Available online 15 August 2012 Keywords: Copper removal Activated unsaturated sand filter Manganese Electrostatic attraction Precipitation Adsorption

C/Cin

Electrostac

0.8

Precipitaon

0.6

Copper/Mn-coated sand

0.4

Adsorpon/ion exchange

0.2 0 0

a r t i c l e

Mn-coated sand

Sand

C/Cin,q/qsat

environment and can have several health effects. " Activated unsaturated sand filter was effective for Fe and Mn but not studied for Cu. " AUSF was indeed found effective for removing copper from water. " Removal mechanisms were suggested and supported by experiments.

10

20

30 40 me (min)

50

60

a b s t r a c t Activated unsaturated sand filter (AUSF) is one among the few of the filtration technologies utilised to treat metal bearing waters. AUSF utilises activated sand as its filter media and operates under naturally flowing air to ensure unsaturated condition. In this study, manganese-activated sand was used for the removal of copper. Activation of the sand increased its BET surface area by 21% at 0.071 mg Mn/g sand and Mn leaching was significant only at pH < 4.5. SEM/EDS showed clusters of crystalline manganese oxides coating the sand surface and newborn copper hydroxides occupying sand surface after exposure to copper-bearing-water. A tracer study revealed that the gas voidage in the filter increased from 0.143 to 0.294 as the water flow rate reduced from 74.4 to 16.4 mL/min, which ensured better water aeration and increased copper removal. The ratio of manganese to sand was a key parameter as it increased almost exponentially the removal capacity of the AUSF. On the other hand, copper removal increased linearly with the inverse of the sand particle size, which indicates the importance of the surface area in the process. However, unexpectedly, when the inlet copper concentration increased from 3 to 20 mg/L, the filter capacity reduced by about 50% from 0.020 to 0.011 mg Cu/g sand. Based on the results obtained, mechanisms by which copper was removed in AUSF were proposed and they were found to vary as function of time [electrostatic attraction (t < 2 min); precipitation (t < 3.8 min), and adsorption/ion exchange (t < 60 min)]. Ó 2012 Elsevier B.V. All rights reserved.

1. Introduction Several research studies have demonstrated the presence of high concentrations of heavy metals in ground waters [1] and surface waters [2,3]. Heavy metals are not biodegradable and tend to accumulate in living organisms causing several health effects. For ⇑ Corresponding author. Tel.: +44 (0) 1792 606841; fax: +44 (0) 1792 295676. E-mail address: [email protected] (C. Tizaoui). 1385-8947/$ - see front matter Ó 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.cej.2012.08.013

instance the presence of copper in water can cause vomiting, diarrhea, stomach cramps, nausea, greenish or bluish stools and saliva and may have severe effect in the brain and liver of people with Wilson’s disease [4,5]. Elevated intake of Cu may cause liver and kidney damage, and sometimes death, particularly in children [4,6,7]. Copper reaches water supplies through discharges from anthropogenic activities such as mining and smelting, industrial activities, municipal wastes and sewage sludge [4,6–8] and also from natural sources such as sea salt sprays, windblown dusts,

335

C. Tizaoui et al. / Chemical Engineering Journal 209 (2012) 334–344

2. Materials and methods All chemicals were purchased from Fisher Scientific, UK, and were of reagent grade. Water solutions containing copper ions (Cu2+) were prepared from copper nitrate (Cu (NO3)23H2O). All aqueous solutions were prepared in Milli-Q water (Q-H2O, Millipore Corp. with resistivity of 18.2 MX cm). All solutions were prepared at initial pH and copper concentrations so that the solubility product of the copper hydroxide (Ksp = 2.35  1020 at 20 °C) was not exceeded to avoid precipitation. The filter media used in this research was natural, uncrushed silica sand fractions B (1.18 mm–600 lm) and C (600 lm–300 lm) (standard sand BS 1881–131:1998 from David Ball Specialist Sand, UK). 2.1. Activated sand preparation First, the sand was sieved to a 0.850 mm diameter fraction using Russell Sieve (Russell Finex, Model 17300). Then the activated sand was produced by drying the sieved sand at 105 °C in an oven for about 12 h then soaked with KMnO4 0.01 N for about 24 h and washed with water before being dried again at 105 °C in an oven for about 12 h [16,17]. The activated sand was then cooled to room temperature and stored in plastic containers before

being used in the experiments. The mass of manganese that effectively coated the sand per unit mass of sand, cMn, was determined by nitric acid digestion using the US EPA method EPA 3050B. 2.2. Analytical Copper concentrations were analysed using Bicinchoninate method (Hach method 8506: spectrophotometer DR-2400, CuVerÒ 1 Copper Reagent Powder Pillow, concentration range 0.04– 5.00 mg/L, wavelength range 400–880 nm, Automatic Wavelength Selection), which is approved by the US Environmental Protection Agency. Samples at higher concentrations than the method range were diluted before analysis. In this method, the copper is determined by the reaction with a salt of bicinchoninic acid (2,2-biquinoline-4,4-dicarboxylic acid) contained in CuVerÒ 1 reagent to form a purple coloured complex in proportion to the copper concentration. The coloured complex is then measured by the spectrophotometer (Hach DR-2400). The accuracy of the method was also checked with a Varian 240FS AAS and the correlation factor R2 was 0.9904. HNO3 (10%) was added to the sample solutions to reduce the pH to less than 6 in order to stabilise the metal ion and prevent precipitation. Point of zero charge of the activated sand was determined by the mass titration method [18]. Briefly, the method involved adding certain masses of sand (0.01%, 0.1%, 1%, 5%, 10%, 20%, 30%, 40%, 50% and 60%) to Milli-Q water in a beaker. This beaker was sealed and placed on a shaker for 24 h. The resulting pH values were measured at the end of the experiment. Acid and alkali resistance tests were also carried out by soaking the manganese-coated sand in Milli-Q water at room temperature for 24 h at pH values adjusted in the range 2–11 ± 0.1 using HNO3 and NaOH [19]. The activated sand was also characterised using photomicrography of its exterior surface acquired by Scanning Electron Microscope (SEM) (Hitachi S-4800, Oxford Instruments). The distribution of elemental concentrations for the sand samples was investigated using the mapping analysis of SEM/Energy Dispersive X-ray Spectroscopy (EDS). Samples for SEM/EDS were placed on a circular carbon film (9 mm diameter) and/or a quick drying silver paint (Agar G302, Agar Scientific, UK) to avoid influence of any charge effect and sample movement during the SEM operation. In addition, the surface area of the prepared sand was determined by the Brunauer, Emmet, Teller (BET) method using nitrogen adsorption at 77.3 K technique (NOVA 2000e Surface Area & Pore Analyzer). 2.3. Activated sand filter A schematic diagram of the activated sand filter is depicted in Fig. 1. The filter used in this research was made of a QVF glass col-

Top liquid distributor

Influent tank

Activated sand

volcanogenic particles and decaying vegetations [4]. The removal of copper from water is hence very important in order to protect water supplies and human health. Precipitation followed by filtration tends to be the most common technique used due to its simplicity [9]. Essentially, precipitation involves changing the dissolved metal ions into insoluble solid state species by a chemical reaction with a precipitant such as alkali or sulphide. It is common that the insoluble solids are removed by filtration, which involves fully submerged filter media in water [10,11]. Integration of the two steps (i.e. precipitation reaction and filtration) in one step seems beneficial. To this end, this can be achieved by filtration on reactive media and such technique is termed as activated unsaturated sand filter (AUSF), which was first studied by Paramarta et al. [12] for the removal of iron and manganese. Basically AUSF is an activated granular filter media (e.g. sand) operating under unsaturated conditions (i.e. water does not fully occupy all the filter pores). The activation of the media can be done by KMnO4 solutions and previous studies have shown that this technique leads to faster reaction rates of metals removal [13– 15]. Moreover, the presence of water-free pores in the filter aerates the water which also promotes the chemical reactions in solution [12]. Solid metal species formed as a result of the precipitation reaction precipitate and are removed on the same filter media. Since this treatment technique combines both precipitation and filtration processes in a single unit, it offers the advantage of not requiring a separate sedimentation unit. AUSF was found an effective alternative for treating iron and manganese from water and was able to achieve concentrations that met for example the Indonesian water standards (Fe: 0.3 mg/L, Mn: 0.1 mg/L) or even below. Besides, AUSF costs were significantly low as compared to those reported by the Indonesian water utility [12,16]. Despite the fact that AUSF technique was efficient to remove Fe and Mn, currently there are only a very limited number of research studies on AUSF removal of other metals and this study is the first of its kind to investigate copper removal from water in AUSF. The performance of AUSF in removing copper at different operating conditions was hence investigated in this study. In addition, the hydraulic performance of AUSF using tracer studies and characterisation of the activated sand were made. Suitable mechanisms for copper removal supported by experimental results were also suggested in this study.

Bottom perforated plate

Pump

Effluent

Fig. 1. Schematic diagram of the AUSF.

336

C. Tizaoui et al. / Chemical Engineering Journal 209 (2012) 334–344

umn of 40 mm internal diameter and 610 mm total height. Manganese-coated sand (MCS) was transferred into the column up to a desired height (generally 450 mm). Water was pumped from a 20 L tank through a 5 mm diameter plastic hose. The filter was equipped with two perforated plates; one was placed at the top of the column and the other at the bottom. These perforated plates were used so as to provide air in the column naturally. The water was introduced at the top of the filter through a perforated plastic sheet in order to distribute water evenly through the cross sectional area of the column. The perforated plate at the bottom was equipped with an aluminium screen (hole diameter: 2 mm, thickness: 75 mm) to support the MCS bed and avoid it from escaping the column. Water samples were collected at the outlet of the filter. To obtain unsaturated condition, starting from dry bed, the input water flow rate (Q) was set so as to avoid flooding the column. On the other hand, to obtain saturated condition (ASSF) the whole manganese sand bed was initially submerged in water before the water flow started. 3. Results and discussion The following definitions have been used in this study. The mass of copper retained in the filter, mr, was calculated using an overall mass balance around the column i.e. the rate of accumulation of copper within the column is equal to the inlet flow rate of copper minus the outlet flow rate of copper. Suppose Q is the flow rate of feed water assumed constant, Cin is the feed concentration, and C is the effluent concentration, the mass balance can then be written as:

QC in  QC ¼

dmr dt

ð1Þ

Integration of Eq. (1) gives the mass of copper removed as function of time, t (Eq. (2)).

mr ¼ QC in

 Z t C dt 1 C in 0

ð2Þ

Based on the experimental data, the mass of copper removed by the activated sand column (mr) at a time t was calculated using Eq. (3), which was derived from Eq. (2) where the trapezoidal rule was used to calculate numerically the integral part [20] – ti is time of the ith observation (min) and ti+1 is time of the ith+1 observation (min). Besides the mass removed by the sand, Eq. (3) accounts for a negligible mass of copper accumulated within the filter voidage as well, which was found less than 4% of the total as calculated by Eq. (3).

mr  QC in

     X C C 0:5  ðt iþ1  ti Þ 2   C C in iþ1 in i i

ð3Þ

The removal capacity of the filter, q, at a particular time was defined as the mass of copper removed, mr, over the mass of sand, ms, (q = mr/ms), and the maximum or saturation capacity of the filter is qsat. In addition, the removal percentage R (%), is defined as R = 100  (Cin  C)/Cin [21] and the empty-bed contact time is defined as EBCT = Vb/Q where Vb is the bed volume [22]. 3.1. Sand characterization 3.1.1. Scanning Electron Microscope (SEM) images SEM images were made to investigate the surface morphology of manganese-coated and uncoated sand and are shown in Fig. 2a–c. Fig. 2a shows that the uncoated sand has a fairly uniform surface with small fractures giving a lightly rough surface. This morphology of sand particles was also observed by other studies

Fig. 2. SEM images for (a) uncoated sand 10,000 magnification (b) fresh coated sand (cMn = 0.071 mg Mn/g sand) 10,000 magnification (c) fresh coated sand (cMn = 0.071 mg Mn/g sand) 40,000 magnification (d) used coated sand (cMn = 0.071 mg Mn/g sand, Cin = 20 mg Cu/L) 5000 magnification.

C. Tizaoui et al. / Chemical Engineering Journal 209 (2012) 334–344

[23,24]. In contrast, the coated sand (Fig. 2b) emerged as a rough surface due to newborn coating layers, possibly of manganese oxides. Fig. 2b also shows no visible uniform sites apart from the potential crystalline manganese oxides, formed in clusters, and appeared on occupied surfaces. These clusters are clearly shown through higher magnification of 40,000 (Fig. 2c). Hu et al. [19], Han et al. [23], Lee et al. [25] and Tiwari et al. [21] have also shown similar clusters as those presented in Fig. 2c. Fig. 2d shows an SEM image of the coated sand that was used to treat copper bearing water. The coated sand surfaces were apparently occupied by newborn copper hydroxides and complexes in addition to manganese oxides. This SEM result suggests that copper may well be removed by precipitation and adsorption on the surface of the manganesecoated sand. 3.1.2. SEM/EDS analysis SEM/EDS analysis was made to determine and confirm the elements on fresh and used manganese-coated and uncoated sand. The peak heights in the EDS spectra are proportional to the amounts of the metallic elements. The energy spectrum for uncoated sand is shown in Fig. 3a. High level of silica (Si) and oxygen can be observed as these are the principal elements of quartz sand. This is confirmed by other studies [21]. Other elements including Al, Fe, K, Mg and Ca were also detected. Knowing that the sand used in this study originates from Lower Greensand (Leighton Buzzard, Beds, UK) [26], it is likely that the detected Al comes from glauconite [27], which is occasionally detected in the Lower Greensand. Iron, calcium, magnesium and potassium were slightly observed as they are also components of glauconite [27]. In addition, sulphur and carbon were slightly detected as also re-

337

ported by Lee et al. [28]. Other studies also found Fe, Al, Mg and K signals in the raw sand they used [21]. The energy spectrum for fresh MCS is shown in Fig. 3b. Besides the signals described above for the uncoated sand, manganese signals were clearly observed, which proves the attachment of manganese species to the surface of sand. The figure is in good agreement with other studies [19,23]. Since a strong peak of Si is observed in the energy spectrum of MCS, it is clear that manganese did not coat the entire surface of sand. Moreover, the EDS elemental distribution mapping of Si and Mn on MCS is illustrated in Fig. 3c. Bright points correspond to the signal of the element from the coated sand. The figure reveals that manganese was concentrated only on certain areas of the sand surface, its distribution was not uniform, and at lower amount than Si. This gives evidence that manganese effectively coated the sand as is also clearly shown by the SEM/EDS line scanning (Fig. 3d). The cluster shown in Fig. 3d, which results from manganese as clearly evidenced by its peak (Fig. 3d inset), gives an insight into the way manganese is attached to the sand surface and confirms the non-uniformity of manganese distribution on the surface. A mechanical interfacial layer between manganese and sand was assumed as a mean for attachment of manganese to the sand surface [19]. The energy spectrum of the used MCS for copper removal is shown in Fig. 3e. Besides all elements discussed above, Fig. 3e clearly shows the appearance of copper signals. This indicates that copper was effectively removed on the surface of MCS possibly through precipitation and adsorption mechanisms. Moreover, SEM/EDS line scanning (results are not shown) shows a noticeable level of copper alongside manganese within a cluster, proving the attachment of copper to the sand surface. Han et al. [23] have also

Fig. 3. SEM/EDS images for (a) uncoated sand and (b) fresh coated sand, (c) elemental distribution mapping for fresh coated sand, (d) line scanning for fresh coated sand, (e) SEM/EDS of used coated sand – conditions as in Fig. 2.

338

C. Tizaoui et al. / Chemical Engineering Journal 209 (2012) 334–344

2

Surface area (m2/g sand)

Uncoated sand Fresh coated sand (cMn = 0.071 mg Mn/g sand) Used coated sand under saturated condition (Cin = 20 mg Cu/L) Used coated sand under unsaturated condition (Cin = 20 mg Cu/L)

0.302 0.365 0.163 0.084

shown by SEM/EDS, the appearance of copper on their used sand in a batch system. Fig. 3e also shows peaks of manganese indicating that not all manganese sites were occupied (or masked) by copper. Indeed calculation (experimental conditions as in Fig. 2d) showed that only 45% of the Mn active sites were potentially occupied by copper. 3.1.3. Specific surface area analysis Surface areas for sand samples were determined using Brunauer, Emmet, and Teller (BET) analysis. Table 1 lists specific surface areas for uncoated, fresh coated, and used coated sands under saturated and unsaturated conditions. The surface area of sand increased by 21% from 0.302 to 0.365 m2/g sand after manganese coating at 0.071 mg Mn/g sand. Han et al. [23] who used Mn to sand ratio higher than this study (5.46 mg Mn/g sand) also observed an increase in surface area but by only 6%. However, Tiwari et al. [21] who used a lower Mn to sand ratio (1 mg/g sand) reported an increase in surface area of about 29%, which is comparable to this study. In contrast, for the used MCS, the surface area has almost halved (0.163 m2/g sand) from that of fresh MCS, a result which is also in agreement with Han et al. [23]. This reduction in surface area may due to blockage of the area of fresh MCS as a result of adsorption and precipitation of copper. BET surface area measurements have also revealed that the surface area of used MCS under unsaturated condition (0.084 m2/ g sand) is about half that under saturated condition (0.163 m2/ g sand). Owing this reduction to blockage by copper precipitation, this indicates that operating the filter under unsaturated conditions results in better removal of copper. 3.1.4. Acid and alkali resistance tests Fig. 4 shows the results from the acid and alkali resistance tests. These tests are utmost important as they describe the attachment strength between manganese and sand surface. The figure indicates that as the pH decreased below about 4.5, a significant release of manganese into the aqueous solution is observed. Manganese concentrations as high as 1.8 mg/L were measured at pH 2. Hu et al. [19] and Tiwari et al. [21] have also observed dissolution of manganese from the Mn-coated sand at pH 2. Higher pH values than 4.5 gave almost no release of manganese into solution. This clearly indicates that pH affects the attachment of manganese on the sand surface and a careful control of pH higher than 4.5 assures high resistance to manganese leaching. 3.2. Hydraulic performance Tracer studies were carried out to determine the hydraulic performance of the AUSF. A pulse of 10 mL at 100 mg/L sodium chloride, a common tracer [22], was injected into the influent end of the AUSF. From the studies, the theoretical mean detention time, tc , the spread of the distribution measured by the variance r2 c and the total mass of salt measured at the reactor effluent, msalt were determined [22]. The tracer experiments were carried out at various flow rates (74.4, 55.0, 37.8, and 16.4 mL/min) but at

1.6 1.2 0.8 0.4 0

0

2

4

6 pH

8

10

12

Fig. 4. Acid and alkali resistance tests (dsand = 0.850 mm, cMn = 0.071 mg Mn/g sand).

40

Na Cl concentration (mg/L)

Sand type

Mn concentration (mg/L)

Table 1 Specific surface area of sand at different conditions (dsand = 0.850 mm).

30

16.39 mL/min 37.78 mL/min

20

55.01 mL/min 74.40 mL/min

10

0 0

2

4 6 Time (min)

8

10

Fig. 5. C-curves (msalt = 1 mg; dsand = 0.850 mm, msand = 892.28 g, H = 45 cm).

fixed mass of sand of 892.28 g and bed height of 45 cm. The ccurves obtained at these different flow rates are shown in Fig. 5 and their corresponding residence times and variances are summarized in Table 2. Consistency checks were made and showed that most of the tracer was recovered by more than 95% on average at the effluent of the column, which validates the study. Fig. 5 shows non symmetrical c-curves with end–tails. This indicates that the flow approaches plug flow but with dispersion. The degree of dispersion is characterised by calculating the dispersion number (dn = D/uL) from r2 using the open–open vessel model 2 ðr2 ¼ 2dn þ 8dn Þ [29]. An average value of dn was obtained = 0.041 ± 0.016. The magnitude of dn number value indicates that the flow did not deviate largely from plug flow, which validates the model used to describe the observed experimental results. Results of the variance (Table 2) show that the spread of the data from the mean is not significant (average 5%), hence the mean residence time from the c-curves, t c , can be defined as the ratio of VL/Q, where VL represents the volume occupied by the liquid in AUSF and Q is the flow rate of water passing through it. The fraction of volume occupied by liquid to the total column volume is eL = VL/V and the fraction of volume occupied by the gas under experiment to the total column volume is eG = e  eL, where e is the total voidage in the column having a value equal to 0.38 determined in a preliminary experiment. Values of eG are shown in Table 2. It is clear from the results in Table 2 that as the flow rate increased, the fraction of gas available in the column decreased since more liquid occupied more of the column voidage. The higher the eG (i.e. low flow rate), the better is aeration of the water, which would potentially result in better removal of metals. Moreover, the

339

C. Tizaoui et al. / Chemical Engineering Journal 209 (2012) 334–344 Table 2 Hydraulic performances derived from tracer studies (msalt = 1 mg; dsand = 0.850 mm, msand = 892.28 g, H = 45 cm). Flow rate (mL/min) Mean residence time from c-curve (min) Variance (min2)

eG

16.4 2.95 1.06 0.294

37.8 2.29 0.83 0.227

55 1.99 0.43 0.186

74.4 1.8 0.42 0.143

results clearly prove that when the flow was introduced at the top of the column, the sand bed was not fully submerged in water, hence the system operated under unsaturated conditions.

3.3. Copper removal under water saturated and unsaturated flow conditions The outlet copper concentration was measured as function of time and the mass of copper removed at a time t was calculated (Eq. (3)). Fig. 6a shows typical breakthrough curves for copper removal on the filter operated under saturated (ASSF) and unsaturated (AUSF) conditions. The figure shows that the outlet concentrations followed similar trend; being low initially, indicating copper removal, and evolved until reaching the inlet concentration, Cin. The figure shows that the profile of the c-curves do not follow a smooth trend possibly due to changing removal mechanisms over time as discussed in the proposed mechanisms. The calculated removal capacity, q, evolved as function of time to a maximum (or saturation) value noted as qsat. Fig. 6a and b clearly shows that AUSF was more efficient than ASSF in removing copper. As shown in Fig. 6a, ASSF reached saturation in about 50 min while AUSF continued removing copper until about 80 min. This indicates that AUSF can operate much longer than ASSF. The removal percentages obtained in this study were lower than those obtained by Lee et al. [28] who reported removals up to 58.2% and 71.2% for non-aerated and aerated conditions respectively. This is likely due to the low Mn to sand ratio used in this study (0.071 mg Mn/g sand) as compared to that of Lee et al. [28] (8.03 mg Mn/g sand). Moreover and possibly for the same reason, saturation times were shorter in this study as compared to 1000 min [28]. The removal percentage of copper (for unsaturated condition) after 60 min was 14.9% which was also lower than the 80.0% obtained by Han et al. [23]. This again is likely due to the higher manganese ratio of 5.46 mg Mn/g sand. Previous studies showed that complete removal of copper and higher saturation times were obtained when

(a)

1

smaller particle sizes [30] or different material (i.e. immobilised biopolymers [31] and chitosan coated sand [24]) were used. Fig. 6b also shows that qsat achieved in ASSF was 0.0058 mg/g as compared to almost a double value of 0.0106 mg/g in AUSF. This indicates that unsaturated condition (i.e. aerated) enhanced the filter performance, which is in agreement with Lee et al. [28] who have also obtained enhanced removals under aerated conditions. However, the qsat values obtained in this study were lower than that of Lee et al. [28] (0.70 mg copper/g MCS) since they used more manganese to coat the sand particles. It is likely that the enhanced removals under aerated conditions were due to the formation of manganese oxide as a result of manganese reaction with oxygen. Knowing that manganese oxide has a pHpzc value of 3 [32], hence at the experimental pH, which was higher than 3, its negatively charged surface attracts the positively charged copper ions facilitating their removal from solution. Although, it is general that pH adjustments are made in order to achieve removal of the metal hydroxide by precipitation using chemicals such as lime or soda caustic [24,28], which can have drastic effect on the overall cost of the operation, in AUSF such pH adjustment is not needed, hence its economical benefit. 3.4. Effect of manganese to sand ratio, cMn, on AUSF performances The manganese to sand ratio, cMn, was determined by acid digestion and its effect on copper removal was studied – results are depicted in Fig. 7a and b. Height of sand bed (H), flow rate (Q), initial concentration (Cin) and sand diameter particle size (dsand) were kept constant at 450 mm, 80.69 mL/min, 20 mg/L and 0.710 mm respectively. Fig. 7a shows that q increased as cMn increased. This can be explained by increased active sites for copper removal as more manganese coated the sand particles. Moreover as shown in Fig. 7b and for the same reason, the maximum copper removed qsat enhanced significantly as cMn increased. An exponential function seems to fit well the change of qsat as function of cMn. On a molar basis, the amount of copper removed at saturation relative to manganese can be calculated by the ratio (MMn  qsat)/(MCu  cMn) where MMn and MCu are the molecular masses of Mn and Cu respectively. This ratio varied between almost stoichiometric ratio of 1.0 up to 2.7 mol Cu/mol Mn as cMn increased from 0.071 to 0.134 mg/g. This indicates that at lowcMn copper removal is based on the interaction with the manganese species possibly by adsorption/ion exchange on a molar basis but ascMn increased, additional mechanisms such as precipitation and/or multilayer adsorption would contribute to the overall removal process.

(b)

0.012 0.01

0.8

0.4

AUSF

ASSF

0.2

q(mg/g)

C/Cin

0.008 0.6

0.006 0.004 AUSF

0.002

ASSF

0

0 0

20

40 60 Time (min)

80

100

0

20

40 60 Time (min)

80

100

Fig. 6. Copper removal under saturated (ASSF) and unsaturated (AUSF) conditions (Cin = 5 mg/L, Q = 81.9 mL/min, dsand = 0.850 mm, H = 45 cm, cMn = 0.071 mg Mn/g sand) (a) effluent copper concentration, C/Cin vs. time (b) removal capacity, q vs. time.

340

C. Tizaoui et al. / Chemical Engineering Journal 209 (2012) 334–344

(a)

0.134 mg Mn/g

0.5

(b) 0.5

0.126 mg Mn/g 0.071 mg Mn/g

q sat = 0.0092×exp(26.76γ Mn ) R² = 0.98

0.4 qsat (mg Cu/g sand)

q(mg Cu/g sand)

0.4

0.000 mg Mn/g 0.3

0.2

0.1

0.3

0.2

0.1

0

0 0

100

200 time (min)

300

400

0

0.05 0.1 Mn to sand ratio, γ Mn (mg/g)

0.15

Fig. 7. Effect of manganese to sand ratio, cMn, on (a) removal capacity change with time, and (b) saturation removal capacity, qsat, (Cin = 20 mg/L, Q = 80.69 mL/min, dsand = 0.710 mm, H = 45 cm).

3.5. Effect of height of sand bed on AUSF performances The effect of height of sand bed (H = 150, 250, 350 and 450 mm) on copper removal was studied at constant values of Cin, Q, dsand and cMn set at 20 mg/L, 80.91 mL/min, 0.85 mm and 0.071 mg/g respectively. The desired bed heights were set using different masses of sand that varied from 320.28 g to 892.28 g. The results show that the curves representing the change of copper concentration ratio (C/Cin) as function of the number of equivalent bed contact time (i.e. Q  t/Vbed) had the same shape and almost superimposed (not shown here) which indicates, under the conditions studied, that the height of sand does not affect the mechanisms by which copper is removed. Other researchers found that C/Cin vs. time (t) curves were steeper at lower bed depth [33–35]. Moreover, the sand saturation capacity, qsat, was calculated and it was found not to vary with sand height. A value of 0.0098 ± 3% mg Cu/g sand was determined. 3.6. Effect of copper input concentration, Cin, on AUSF performance The effect of inlet copper concentration, Cin, varying between 3 and 20 mg/L on copper removal was studied and the results are shown in Fig. 8a and b. Fig. 8a shows that at low Cin the breakthrough curves extend for longer times as compared to those at higher Cin, which are characterised by a sharp increase in concentration possibly due to low ratio of active sites to the amount of copper introduced to the filter. Other researchers also found similar tendency [31,33,34,36]. The removal capacity, q, evolved as function of time to qsat and the change of qsat as function of the inlet copper concentration is shown in Fig. 8b. Surprisingly, the figure shows that as the inlet concentration increased, the saturation removal capacity of AUSF decreased. Experimental results showed that at all inlet copper concentrations, the pH of the solution exiting the filter increases initially to a high value at around 8.19 ± 0.11 then decreases to a value at about 5.51 ± 0.04 (Fig. 8c – for clarity, results of only three inlet concentrations were plotted). On one hand, Fig. 8c shows that the rate of pH decrease was faster at high Cin due to a rapid consumption of hydroxide ions, which results in rapid copper precipitation and on the other hand, Fig. 8d shows clearly that at high Cin, the removal of copper at initial times, which is essentially by precipitation since the ion product exceeded the solubility product, was significantly high than that at low Cin. This suggests that at high Cin, significant mass of precipitates coat the surface of MCS, which has potential to block the removal sites and hence reduces the overall capacity of the fil-

ter. In contrast and for dilute concentrations, precipitation is less significant in relation to the available sites, hence active sites for adsorption are more available for copper removal. The low change in pH observed in Fig. 8c at low Cin suggests that copper removal by adsorption dominates at low Cin. Breakthrough curves obtained at low inlet concentrations as opposed to those obtained at high inlet concentrations were steep at the beginning but concave downward with increasing operation time and asymptotically nearing complete breakthrough. This type of breakthrough curve is typical for adsorption where solid-phase mass transfer controls the process, which further supports that copper was removed by adsorption at low inlet concentrations. 3.7. Effect of liquid flow rates on AUSF performances The effect of liquid flow rate at values varying between 37.78 and 81.91 mL/min on AUSF copper removal was studied. Results show that C/Cin vs. time (t) curves were steeper at high flow rates (not shown here). This occurs as the greater the flow rate, the more void occupied by the aqueous solution thus saturation is achieved faster, which is in agreement with other studies [31,33,34,36]. The change of the ratio q/qsat as function of the number of equivalent bed contact time is represented in Fig. 9a at the different flow rates. The figure shows that all curves almost superimpose, particularly for q/qsat < 0.8. This indicates that liquid-film mass transfer is not the controlling process, but as reported earlier, solid-phase mass transfer controls the process. The change of qsat as function of liquid flow rate is represented in Fig. 9b. The figure shows that qsat reduces by about 10% as the liquid flow rate increases from 37.78 to 81.91 mL/min. The reason for this reduction may be explained by the reduction in air fraction due to increased liquid flow rate (Table 2); a clear relationship was obtained between reduction in qsat and eG as the flow rate increased (Fig. 9b). This indicates that aeration plays a significant role in copper removal by AUSF. 3.8. Effect of sand particle diameters on AUSF performance The effect of sand particle diameter on copper removal was studied and the results are shown in Fig. 10. For particle diameter of 0.4 mm, the liquid flow rate was reduced from 80.92 mL/min to 25 mL/min to overcome the high pressure drop exhibited at the high flow rate. Fig. 10 depicts the change of q/qsat as function of the number of EBCT. Results show that complete removals of copper and higher saturation times were obtained at smaller sand sizes, which is in agreement with Boujelben et al. [30]. Although

341

C. Tizaoui et al. / Chemical Engineering Journal 209 (2012) 334–344

(a)

(b)

1

2.0

0.8

qsat(mg/g sand)

3 mg/L 4 mg/L

C/Cin

0.6

5 mg/L 10 mg/L

0.4

15 mg/L

1.5

1.0

0.5

0.2

20 mg/L 0.0

0 0

50

100 time (min)

150

0

200

5

10 Cin (mg/L)

15

20

0.012

9

(c)

2.5

(d)

pH of solution exiting filter

0.01 q(mg Cu/g sand)

8

Cin=3mg/L Cin=5mg/L 7

Cin=20mg/L

0.008 0.006 0.004

6

Cin=20 mg/L Cin=5 mg/L Cin=3mg/L

0.002 5

0 0

5

time (min)

10 time (min)

15

20

Fig. 8. Effect of copper inlet concentration on copper removal by AUSF (Q = 80.91 mL/min, dsand = 0.850 mm, H = 45 cm, cMn = 0.071 mg Mn/g sand) (a) change of C/Cin as function of time; (b) variation of AUSF saturation removal capacity as function of inlet copper concentration; (c) change of effluent pH as function of inlet concentrations; (d) effect of Cin on q at initial times of the experiments.

1

(b)

1.36

0.8

100×qsat (mg/g)

0.6

55.01 mL/min 74.40 mL/min

0.4

0.22

1.34

37.78 mL/min

q/qsat

0.24

1.38

81.91 mL/min

0.2

0.2

1.32 1.30

0.18

1.28

0.16

1.26

qsat qsat

0.14

εG eG

0.12

1.24 1.22 1.20

0 0

5 10 Number of EBCT

15

gas fraction, ε G

(a)

0.1 20

50 80 Q (mL/min)

110

Fig. 9. AUSF performances at different flow rates (Cin = 5 mg/L, dsand = 0.850 mm, H = 45 cm, cMn = 0.071 mg Mn/g sand) (a) q/qsat vs. number of EBCT, (b) variation of qsat and gas fraction eG as function of liquid flow rate.

the curve profiles shown in Fig. 10 are similar, it is clear that as the particle diameter reduced, the curves became broader as the operation time increased. Moreover, Fig. 10 inset shows that qsat increased almost linearly as the inverse of the particle size increased (i.e. surface area increased), which indicates that surface area plays an important role in copper removal.

4. Copper removal mechanisms Based on the results obtained in this study, the mechanisms underpinning copper removal have been proposed. The chemical reactions for copper removal in AUSF are somewhat different from those for iron and manganese since the trivalent copper is unstable

C. Tizaoui et al. / Chemical Engineering Journal 209 (2012) 334–344

1

(a)

0.85 mm 0.71 mm 0.5mm 0.4mm

0.8

7 2.5

pH

2 1.5

0.4

0.8

6

×

q/qsat

1

8

0.6

5

0.6 pH input pure water pH effluent pure water pH input Cu solution pH effluent Cu solution pHpzc C/Cin q/qsat

4 3

1

0.2

9

2

0.5 1 1.5 2 2.5 3

1

0.4 0.2

0

0 0

5 10 number of EBCT

0 0

15

Fig. 10. Effect of sand particle diameter on copper removal (Cin = 5 mg/L, H = 45 cm, cMn = 0.071 mg Mn/g sand).

(b)

C/Cin or q/qsat

342

10

20

30 40 Time (min)

50

60

100

Cu2+ 80

þ þ ½CuðH2 OÞ6 2þ ðaqÞ þ H2 OðlÞ ¢ ½CuðH2 OÞ5 OHðaqÞ þ H3 OðaqÞ

ð4Þ

4.1. Precipitation As evidenced by experiments in this work, MCS releases hydroxide ions in the aqueous solution leading to increased pH. Such conditions are expected favourable for precipitation reactions to take place in which hydroxide ions react with the hexaaquacopper complex in two steps (Eqs. (5) and (6)) to form the solid precipitate and this results in the removal of ionic copper from solution. The overall precipitation reaction is represented by Eq. (7) or sim-

Species (%)

Cu(OH)2 60

Ctotal = 20mg/L Ctotal = 5mg/L

40 20 0 4

(c) pH at which precipitation starts

and hence rarely formed naturally [4]. Early on the experiment, the pH of input copper solution was weakly acidic (about 5.59), which is typical for 2+ metal ions (Fig. 11a). As copper is dissolved in water, it forms the complex hexaaquacopper(II) ion (Cu(H2O)6)2+ in which water is the ligand [37] and due to the pull of the water’s OAH bond electrons towards the positive central ion, the hydrogen atoms attached to the water ligands become sufficiently positive that they can be detached in a reaction involving water molecules to produce H3O+, hence a drop of pH (Eq. (4)). On the other hand, as evidenced by experiments in this work, the manganese-coated sand releases hydroxide ions in the aqueous solution (Fig. 11a – increase of effluent pH as compared to the input pH). In addition, Han et al. [23] and Chang et al. [38] proved that manganese coated on the surface of sand takes predominantly the form of MnO2. Based on a value of Ksp = 2.35  1020 at 20 °C, Fig. 11b shows the speciation curves for copper at two total copper concentrations of 20 and 5 mg/L – other copper species than Cu2+ and Cu(OH)2 (i.e.  2 + Cu2 ðOHÞ2þ 2 ; CuðOHÞ3 ; CuðOHÞ4 , Cu(OH) ) were neglected since the values of the equilibrium constants of the reactions leading to their formation are very low [21,39]. The figure shows that as the total concentration of copper reduces from 20 to 5 mg/L, the critical pH at which precipitation starts to occur increases from about 5.92 to 6.23. At these concentrations, for pHs below the critical values, the copper exists predominantly in its ionic form Cu2+ and as the pH increases to about 7.09 and 7.39 respectively copper is totally converted to precipitates (i.e. Cu(OH)2). Lee et al. [28] also stated that the copper species found in the solution were those of Cu2+ (the water ligand is not shown) once the pH of the solution was below 6.01 for their initial copper concentration of 30 mg/L at 25 °C. The observations outlined earlier are important to elucidate the mechanisms involved in copper removal by AUSF, which are discussed below and are expected to change over time as a result of pH and concentration changes in the filter.

5

6 pH

7

8

7.2 7.0

Ksp = 2.35×10-20 at 20o C

6.8 6.6 6.4 6.2 6.0 5.8 0

5

10

15

20

Fig. 11. Copper removal (a) Changes of pH and concentration and filter capacity ratios as function of time (Cin = 20 mg/L, Q = 80.92 mL/min, dsand = 0.850 mm, H = 45 cm, cMn = 0.071 mg Mn/g sand), (b) copper speciation at 20 °C, (c) theoretical pH at which precipitation occurs.

ply by Eq. (8). The release of hydroxide ions is in line with the experiment carried out with pure water. Indeed the pH of influent pure water was about 6.15 and after passing through AUSF, this pH increased to 8.35 and as water passed through the column, the pH gradually reduced to 6.47 in 60 min (Fig. 11a). þ  ½CuðH2 OÞ6 2þ ðaqÞ þ OHðaqÞ ¢ ½CuðH2 OÞ5 OHðaqÞ þ H2 O

ð5Þ

½CuðH2 OÞ5 OHþðaqÞ þ OHðaqÞ ¢ ½CuðH2 OÞ4 ðOHÞ2 ðsÞ þ H2 O

ð6Þ

 ½CuðH2 OÞ6 2þ ðaqÞ þ 2OHðaqÞ ¢ ½CuðH2 OÞ4 ðOHÞ2 ðsÞ þ 2H2 O

ð7Þ

343

C. Tizaoui et al. / Chemical Engineering Journal 209 (2012) 334–344  or simply : Cu2þ ðaqÞ þ 2OHðaqÞ ¢ CuðOHÞ2ðsÞ

K sp ¼ 2:35  1020 at 20  C

0

2 min

Electrostatic attraction

ð8Þ

The precipitation reaction proceeds up to a point at which the ion product is less than the solubility product constant (Ksp = 2.35  1020 at 20 °C). Based on effluent concentration and pH (Fig. 11), calculation shows that precipitation proceeded up to time 3.8 min, which corresponds to the time at which effluent copper concentration started to increase from zero (i.e. breakthrough point). Fig. 11c shows that as the concentration of copper increases, the theoretical pH at which precipitation starts to occur decreases. The figure also shows that for a 20 mg/L copper, the pH at which precipitation occurs is about 5.92 and since during experiment, the pH for the first 3.8 min was in a range higher than 5.92 (Fig. 11a), precipitation is assumed to be the dominant process. As shown in Fig. 11a, the removal capacity of AUSF within this period (i.e. 0 < t < 3.8 min) is about 20% of the saturation removal capacity, qsat. Following this period of the experiment, the pH then reduced at a very slow rate from 5.92 to about 5.40 in over 50 min. Other studies reported that the precipitation of Cu(OH)2 coats the sand surface forming surface complexes by which copper ions are bound to the sorbent material [23,28,40,41]. Moreover, based on effluent pH change, it was found that the removed copper was significantly higher than the stoichiometric ratio of 2 (Eq. (8)), which indicates that other mechanisms besides precipitation contributed to copper removal. 4.2. Electrostatic attraction and adsorption Adsorption due to electrostatic attraction results from the interaction between species having opposite charges and since the charge of the sorbent surface is affected by pH, the pH of point of zero charge (pHpzc) of MCS was determined in this study. A value of pHpzc equal to 7.78 was obtained, which is within the range of pHpzc values of similar sorbents reported in the literature [21,25,39]. At pH values above the pHpzc, the surface of MCS is negative (i.e. „MnO predominates; Eq. (9)), which favours the removal of any positively charged hexaaquacopper complex; while at pH values less than pHpzc, the removal by electrostatic attraction is expected to be less significant since positive surface predominates (i.e. MnOHþ 2 ; Eq. (10)). As shown in Fig. 11a, the pH of the effluent copper solution dropped below pHpzc rapidly within the first 2 min indicating that electrostatic attraction may contribute to copper removal only during this period of the experiment. The electrostatic attraction results in copper adsorption as illustrated by Eqs. (11) and (12). Adsorption of copper may also take place on a neutral surface as illustrated by Eq. (13).

0

3.8 min

0

Precipitation 60 min

0

0.02

0.04

1.00

0

0.10

0.20

1.00

Adsorption/Ion exchange

C/Cin q/qsat

Fig. 12. Mechanisms contributing to copper removal by AUSF (Cin = 20 mg/L, Q = 80.92 mL/min, dsand = 0.850 mm, H = 45 cm, cMn = 0.071 mg Mn/g sand).

4.3. Ion exchange, adsorption and surface complex formation A further mechanism by which copper is removed from solution may involve ion exchange of H+ with Cu2+ with surface complex formation as illustrated by Eqs. (14)–(16). The release of hydrogen ions resulting from copper ion exchange is evidenced by the more significant drop in pH in the case of copper solution as compared to pure water (Fig. 11a). The decrease in pH values of the copper solution reported in Fig. 11a agrees with the proposed ion exchange mechanism. The figure shows that the decrease in pH took place up to the end of experiment though at very slow rate as compared to the initial part of the experiment (i.e. the first 5 min). This indicates that ion exchange and adsorption processes extend further than precipitation possibly until the end of the experiment.

BMnOH þ Cu2þ ¢ BMnOCuþ þ Hþ

ð14Þ

2ðBMnOHÞ þ Cu2þ ¢ ðBMnO Þ2 Cu2þ þ 2Hþ

ð15Þ

BMnOH þ Cu2þ þ H2 O ¢ BMnOCuOH þ 2Hþ

ð16Þ

BMnOH þ Hþ ¢ BMnOHþ2

ð10Þ

From this analysis, it can be suggested that all mechanisms (i.e. precipitation, electrostatic attraction and ion exchange/adsorption accompanied by surface complexation) contributed to copper removal within the first 2 min, while removal by precipitation continued up to 3.8 min and removal by ion exchange and to some extent adsorption developed up to the end of the experiment (i.e. complete exhaustion of the bed). Precipitation and electrostatic attraction seem to be more significant than the other processes since breakthrough occurred at about the same time when they are inhibited (Fig. 11). A summary of the contribution of each mechanism during the time of the experiment alongside the change of the ratios C/Cin and q/qsat is illustrated in Fig. 12, which shows that about 20% of the total removed copper occurred within the first 3.8 min.

BMnO þ Cu2þ ¢ BMnOCuþ

ð11Þ

5. Conclusions

2ðBMnO Þ þ Cu2þ ¢ ðBMnO Þ2 Cu2þ

ð12Þ

BMnOH þ Cu2þ ¢ CuðBMnOHÞ2þ

ð13Þ

In this study a successful manganese-activated unsaturated sand filter (AUSF) was developed for the removal of copper ions from water. It was found that activation of the sand increased its BET surface area by 21% at a manganese ratio of 0.071 mg Mn/g sand and the leaching of Mn was significant only at pH values less than 4.5. SEM/EDS showed that clusters of crystalline manganese oxides coated the surface of the sand and newborn copper hydroxides and its complexes occupied the surface of sand after exposure to copper bearing water. A tracer study revealed that the gas voidage in the filter increased as the water flow rate reduced, which ensured better aeration of the water and increased removal capacity of the filter. Copper removal was highest in unsaturated acti-

BMnOH ¢ BMnO þ Hþ

ð9Þ

The increase in metal removal as the pH increases (pH = 8.28– 7.00, 0 6 t < 2 min) can thus be explained on the basis of a decrease in competition between proton and copper cations for the same surface functional groups and by the increase in negatively charged surface, which results in a greater electrostatic attraction between the surface and the copper ions. Lee et al. [28] also showed that positively charged copper was adsorbed by negatively charged MCS through electrostatic attraction.

344

C. Tizaoui et al. / Chemical Engineering Journal 209 (2012) 334–344

vated sand filter as compared to saturated activated sand filter. AUSF was advantageous to the usual metal hydroxide precipitation processes as no pH adjustment was necessary. The manganese to sand ratio increased the saturation capacity, qsat, of the AUSF almost exponentially. However, an increase in copper inlet concentration or sand diameter reduced qsat. This suggests that surface area plays an important role in copper removal. Based on the results obtained, mechanisms by which copper was removed in AUSF were proposed and they were found to vary with time. Acknowledgements Sugihhartati Dj Rachmawati thanks the Government of Indonesia (Directorate General of Higher Education, Ministry of National Education) and National Institute of Technology (ITENAS) for sponsoring her PhD study. She is grateful to Alex M. Lord from Swansea University for his help in conducting SEM/EDS analysis. References [1] M.A. Halim, R.K. Majumder, S.A. Nessa, Y. Hiroshiro, M.J. Uddin, J. Shimada, K. Jinno, Hydrogeochemistry and arsenic contamination of groundwater in the Ganges Delta Plain, Bangladesh, J. Hazard. Mater. 164 (2009) 1335–1345. [2] Z. Wang, G. Liu, Z. Fan, X. Yang, J. Wang, S. Wang, Experimental study on treatment of electroplating wastewater by nanofiltration, J. Membr. Sci. 305 (2007) 185–195. [3] S.-S. Chen, C.-Y. Cheng, C.-W. Li, P.-H. Chai, Y.-M. Chang, Reduction of chromate from electroplating wastewater from pH 1 to 2 using fluidized zero valent iron process, J. Hazard. Mater. 142 (2007) 362–367. [4] R. Eisler, Eisler’s Encyclopedia of Environmentally Hazardous Priority Chemicals, first ed., Elsevier, Amsterdam, 2007. [5] A.L. Bojic, D. Bojic, T. Andjelkovic, Removal of Cu2+ and Zn2+ from model wastewaters by spontaneous reduction-coagulation process in flow conditions, J. Hazard. Mater. 168 (2009) 813–819. [6] S.B. Wang, L. Li, Z.H. Zhu, Solid-state conversion of fly ash to effective adsorbents for Cu removal from wastewater, J. Hazard. Mater. 139 (2007) 254– 259. [7] J. Bouzid, Z. Elouear, M. Ksibi, A. Feki, A. Montiel, A study on removal characteristics of copper from aqueous solution by sewage sludge and pomace ashes, J. Hazard. Mater. 152 (2008) 838–845. [8] L.-C. Zhou, Y.-F. Li, X. Bai, G.-H. Zhao, Use of microorganisms immobilized on composite polyurethane foam to remove Cu(II) from aqueous solution, J. Hazard. Mater. 167 (2009) 1106–1113. [9] Q. Chen, Z. Luo, C. Hills, G. Xue, M. Tyrer, Precipitation of heavy metals from wastewater using simulated flue gas: Sequent additions of fly ash, lime and carbon dioxide, Water Res. 43 (2009) 2605–2614. [10] P.d.i.L. Huisman, Mechanical Filtration, Delft University of Technology, 2004. [11] R. Devi, E. Alemayehu, V. Singh, A. Kumar, E. Mengistie, Removal of fluoride, arsenic and coliform bacteria by modified homemade filter media from drinking water, Bioresour. Technol. 99 (2008) 2269–2274. [12] R.B. Paramarta, J.N.I. Wardhana, I. Sodikin, M. Gana, D. Budianto, L.M. Suryodipuro, T. Widayati, Activated un-saturated sand filter as an alternative to reduce Fe & Mn in water treatment [Saringan pasir kering aktif sebagai alternative untuk menurunkan Fe & Mn dalam pengolahan air minum], Bogor, Indonesia, 1988. [13] X. Guan, J. Ma, H. Dong, L. Jiang, Removal of arsenic from water: Effect of calcium ions on As(III) removal in the KMnO4-Fe(II) process, Water Res. 43 (2009) 5119–5128. [14] S. Kawamura, Integrated Design and Operation of Water Treatment Facilities, John Wiley & Sons, Canada, 2000. [15] D. Ellis, C. Bouchard, G. Lantagne, Removal of iron and manganese from groundwater by oxidation and microfiltration, Desalination 130 (2000) 255– 264.

[16] S.D. Rachmawati, D.N. Malasari, M.R. Sururi, A Stratified Activated Dry Sand Filter as an Alternative to Remove Fe and Mn Concentrations in Ground Water Treatment Technology [Saringan pasir kering aktif stratified untuk menurunkan konsentrasi Fe & Mn dalam pengolahan air tanah], Bandung Institute of Technology, Bandung, West Java, Indonesia, 2006. [17] Z. Teng, J.Y. Huang, K. Fujita, S. Takizawa, Manganese removal by hollow fiber micro-filter: membrane separation for drinking water, Desalination 139 (2001) 411–418. [18] J.S. Noh, J.A. Schwarz, Estimation of the point of zero charge of simple oxides by mass titration, J. Colloid Interface Sci. 130 (1989) 157–164. [19] P.-Y. Hu, Y.-H. Hsieh, J.-C. Chen, C.-Y. Chang, Characteristics of manganesecoated sand using SEM and EDAX analysis, J. Colloid Interface Sci. 272 (2004) 308–313. [20] A. Kaw, M. Keteltas, Trapezoidal Rule, University of South Florida, Florida, 2009. [21] D. Tiwari, C. Laldanwngliana, C.-H. Choi, S.M. Lee, Manganese-modified natural sand in the remediation of aquatic environment contaminated with heavy metal toxic ions, Chem. Eng. J. 171 (2011) 958–966. [22] G. Tchobanoglous, F.L. Burton, H.D. Stensel, Wastewater Engineering Treatment and Reuse, fourth ed., McGraw-Hill, Boston, 2004. [23] R. Han, W. Zou, Z. Zhang, J. Shi, J. Yang, Removal of copper(II) and lead(II) from aqueous solution by manganese oxide coated sand: I. Characterization and kinetic study, J. Hazard. Mater. 137 (2006) 384–395. [24] M.-W. Wan, C.-C. Kan, B.D. Rogel, M.L.P. Dalida, Adsorption of copper (II) and lead (II) ions from aqueous solution on chitosan-coated sand, Carbohydr. Polym. 80 (2010) 891–899. [25] S.-M. Lee, W.-G. Kim, C. Laldanwngliana, D. Tiwari, Removal behavior of surface modified sand for Cd(II) and Cr(VI) from aqueous solutions, J. Chem. Eng. Data 55 (2010) 3089–3094. [26] DavidBallSpecialistSands, Standard Reference Materials BS 1881–131: 1998, 2009. [27] G.R. Thompson, J. Hower, The mineralogy of glauconite, Clays Clay Miner. 23 (1975) 289–300. [28] C.I. Lee, W.F. Yang, C.I. Hsieh, Removal of copper(II) by manganese-coated sand in a liquid fluidized-bed reactor, J. Hazard. Mater. 114 (2004) 45–52. [29] O. Levenspiel, Chemical Reaction Engineering, third ed., John Wiley & Sons, New York, 1999. [30] N. Boujelben, J. Bouzid, Z. Elouear, Adsorption of nickel and copper onto natural iron oxide-coated sand from aqueous solutions: study in single and binary systems, J. Hazard. Mater. 163 (2009) 376–382. [31] K.H. Chu, M.A. Hashim, Copper biosorption on immobilized seaweed biomass: column breakthrough characteristics, J. Environ. Sci. 19 (2007) 928–932. [32] M.J. Gray, M.A. Malati, M.W. Rophael, The point of zero charge of manganese dioxides, J. Electroanal. Chem. 89 (1978) 135–140. [33] R.P. Han, W.H. Zou, H.K. Li, Y.H. Li, J. Shi, Copper(II) and lead(II) removal from aqueous solution in fixed-bed columns by manganese oxide coated zeolite, J. Hazard. Mater. 137 (2006) 934–942. [34] S. Kundu, A.K. Gupta, Analysis and modeling of fixed bed column operations on As(V) removal by adsorption onto iron oxide-coated cement (IOCC), J. Colloid Interface Sci. 290 (2005) 52–60. [35] S. Mohan, G. Sreelakshmi, Fixed bed column study for heavy metal removal using phosphate treated rice husk, J. Hazard. Mater. 153 (2008) 75–82. [36] R. Han, L. Zou, X. Zhao, Y. Xu, F. Xu, Y. Li, Y. Wang, Characterization and properties of iron oxide-coated zeolite as adsorbent for removal of copper(II) from solution in fixed bed column, Chem. Eng. J. 149 (2009) 123–131. [37] M.J. Winter, J.E. Andrew, Foundations of Inorganic Chemistry, Oxford Science Publications, New York, 2000. [38] Y.-Y. Chang, K.-H. Song, J.-K. Yang, Removal of As(III) in a column reactor packed with iron-coated sand and manganese-coated sand, J. Hazard. Mater. 150 (2008) 565–572. [39] S.-M. Lee, C. Laldawngliana, D. Tiwari, Iron oxide nano-particles-immobilizedsand material in the treatment of Cu(II), Cd(II) and Pb(II) contaminated waste waters, Chem. Eng. J. 195–196 (2012) 103–111. [40] Z. Reddad, C. Gerente, A. Yves, P.L. Cloirec, Adsorption of several metal ions onto a low-cost biosorbent: kinetic and equilibrium studies, Environ. Sci. Technol. 36 (2002) 2067–2073. [41] C.-I. Lee, W.-F. Yang, C.-S. Chiou, Utilization of water clarifier sludge for copper removal in a liquid fluidized-bed reactor, J. Hazard. Mater. 129 (2006) 58–63.