Efficiency of aluminum-pillared montmorillonite on the removal of cesium and copper from aqueous solutions

Efficiency of aluminum-pillared montmorillonite on the removal of cesium and copper from aqueous solutions

ARTICLE IN PRESS WAT E R R E S E A R C H 41 (2007) 1897 – 1906 Available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/watres ...

391KB Sizes 0 Downloads 12 Views

ARTICLE IN PRESS WAT E R R E S E A R C H

41 (2007) 1897 – 1906

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Efficiency of aluminum-pillared montmorillonite on the removal of cesium and copper from aqueous solutions D. Karamanis, P.A. Assimakopoulos Department of Physics, The University of Ioannina, 45110 Ioannina, Greece

art i cle info

ab st rac t

Article history:

Aluminum-pillared-layered montmorillonites (PILMs) were tested for their potential

Received 7 September 2006

application in the removal of copper or cesium from aqueous solutions. By varying the

Received in revised form

initial conditions, several PILMs were prepared and characterized by means of X-ray

30 January 2007

fluorescence (XRF), proton induced gamma-ray emission (PIGE), X-ray diffraction (XRD) and

Accepted 30 January 2007

sorption isotherms. Uptake of metals was studied by means of XRF spectrometry for copper

Available online 19 March 2007

sorption or g-ray spectrometry for cesium, using 137Cs as radiotracer. The sorption kinetics

Keywords:

and capacity of PILMs were determined in relation to the effects of factors such as the

Pillared clays

initial metal concentration, initial pH of the solution and the presence of competitive

Heavy metals

cations. Kinetic studies showed that an equilibrium time of few minutes was needed for

Copper

the adsorption of metal ions on PILMs. A pseudo-first-order equation was used to describe

Cesium

the sorption process for either copper or cesium. The most effective pH range for the

Montmorillonite

removal of copper and cesium was found to be 4.0–6.0 and 3.0–8.0, respectively. Cesium sorption isotherms were best represented by a two-site Langmuir model while copper isotherms followed the Freundlich or the two-site Langmuir model. Cesium sorption experiments with inorganic or organic competitive cations as blocking agents revealed that the high selective sites of PILMs for cesium sorption (1–2% of total) are surface and edge sites in addition to interlayer exchange sites. In copper sorption, the two sites were determined as interlayer sites of PILMs after restoring their cation exchange capacity and sites associated with the pillar oxides. & 2007 Elsevier Ltd. All rights reserved.

1.

Introduction

In recent years, increased concern of the toxic effects of heavy metal pollution of the environment and particularly bodies of water has resulted in an intensive effort to identify methods for their separation and removal from aqueous solutions. Activated carbon and different types of ionexchange resins are very often used in adsorption processes. However, the high capital and regeneration costs of the materials limits their large-scale use for the removal of metals, and has encouraged researchers to look for low-cost sorbing materials (Bailey et al., 1999; Reddad et al., 2002; Wang et al., 2003; Babel and Kurniawan, 2003). Corresponding author. Tel.: +30 26510 98613; fax: +30 26510 98692.

E-mail address: [email protected] (D. Karamanis). 0043-1354/$ - see front matter & 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2007.01.053

Moreover, one of the effective strategies to address the remediation of a radiocontaminated ecosystem and the management of nuclear wastes is the usage of inorganic ion exchangers as effective trapping agents for the major radiocations (Lin et al., 2001). A number of ion exchangers sorbents (mainly tunnelled and layered structures) are currently being developed for removing 137Cs and/or 90Sr from highly alkaline tank wastes, acid wastes and groundwater (Apak et al., 1996; Solbra et al., 2001; Lahiri et al., 2005; Stout et al., 2006). Although some of these materials have proved effective in special circumstances, a general, efficient, low-cost material for the separation of the two radionuclides of interest is still elusive (Pendelyuk et al., 2005; Gurboga and Tel, 2005; Stout

ARTICLE IN PRESS 1898

WAT E R R E S E A R C H

41 (2007) 1897– 1906

et al., 2006). Consequently, research in the development of effective and low-cost heavy metals and radionuclide sorbents is not only scientifically and economically attractive but also urgent for environmental and health protection. Naturally occurring clay minerals have been extensively studied as sorbents for the removal of various pollutants from wastewater and aqueous solutions (Al-Qunaibit et al., 2005). Montmorillonite, coated and intercalated by aluminum hydroxides, exhibits much higher adsorption capacity for some heavy metal ions than that of natural montmorillonite (Lothenbach et al., 1997). High-temperature calcination of intercalated clays results in ‘‘pillared’’ materials, where the polyhydroxo-cationic species are irreversibly fixed to the layers. Pillared montmorillonites, first tested as cracking catalyst in view of their acidic properties, have recently been studied as sorbents for the removal of hazardous inorganic elements (Karamanis et al., 1997; Cooper et al., 2002; Jiang and Zeng, 2003; Manohar et al., 2005, 2006). However, very little is still known about the removal efficiency and the sorption mechanisms of metals by pillared clays (Manohar et al., 2005, 2006). Therefore, the present work investigated the efficiency of aluminum-pillared-layered montmorillonites (PILMs) for the removal of copper and cesium from aqueous solutions under kinetic and equilibrium conditions. In order to identify the mechanisms of copper and cesium sorption on pillared montmorillonites, the pillaring conditions were varied and different forms of PILMs were prepared and tested in sorption experiments.

2.

Materials and methods

2.1.

Preparation of PILMs

Bentonite was provided by Silver & Baryte Min. Co., Athens. The montmorillonite fraction of the starting bentonite was brought to its Na-homoionic form (Na-montmorillonite, NaM). Specifically, bentonite was repeatedly contacted with 1 M NaCl solutions and washed with deionized water until Cl free (AgNO3 test). Then, a 1 wt% aqueous suspension of the treated bentonite was prepared and left to stand for 24 h. The upper 80% of the colloid column was then separated, centrifuge-washed and finally gently ground after drying at room temperature. An aluminum-pillared montmorillonite, coded Al1MFD3, was prepared according to the method described by Karamanis et al. (1997). Briefly, the Naþ form of montmorillonite was added to the pillaring solution (a 0.2 M solution of AlCl3 with a 0.2 M solution of NaOH up to a molar ratio OH/Al of 1) at a ratio of 3 mmol Al per gram clay and left to react for 1 h under vigorous stirring at 70  C. Once the exchange process was completed, the PILM precursor was repeatedly washed and then freeze-dried. Heating the PILM precursor in air at 500  C for 4 h in order to create the rigid, non-swelling, threedimensional zeolite-like structure (Varma, 2002), delivered the final PILM in powder form. Its cation-exchange capacity was restored by contacting the powder with an ammonia atmosphere and then with a NaCl solution at pH 10 (Karamanis et al., 1997). The PILM material was recovered by centrifugation and was subsequently centrifuge-washed

several times with deionized water; it was finally dried at 80  C. PILMs coded as AlXMFDY were prepared accordingly by varying the OH/Al ratio (X ¼ 0:125, 1 or 2) and the ratio of added mmol Al per gram of clay (Y ¼ 3, 5, 15).

2.2.

Materials characterization

2.2.1.

Elemental analysis

The concentration of each element in the prepared materials was determined with the spectrometric methods of proton induced gamma-ray emission (PIGE) and X-ray fluorescence (XRF) (Karamanis et al., 2001). PIGE measurements were carried out at the 5.5 mV terminal voltage TN11 Tandem accelerator of the National Center for Scientific Research ‘‘Demokritos’’ in Athens. A proton beam with energy of Ep ¼ 4:0 MeV was used and the emitted g-radiation was detected with an 80% high purity germanium (HPGe) detector. Characteristic g-rays emitted from the deexcitation of the residual nuclei following (p,p0 g) reactions, were used for the determination of light elements as Al (1014 keV), Si (1779 keV), Mg (585 keV) and Na (440 keV). For the normalization and extraction of weight percentages, samples of the certified reference material IAEA Soil-7 and of powder graphite mixed with cellulose and a compound containing the element under investigation (in its natural isotopic abundance), were used as standards. XRF measurements were performed at the XRF Unit of the University of Ioannina. A vertical Si(Li) detector was used and the exciting radiation was provided by a ring shaped radioisotope source (109Cd or 241Am). Samples in the form of pellets were placed at the top of the assembly in a pi geometry between the exciting radiation and the samples. Induced X-rays were detected through a small hole in the shielding material. Normalization was performed with direct comparison with the reference IAEA material Soil-7 and matrix effects were corrected through simulation.

2.2.2.

X-ray diffraction (XRD)

XRD patterns of PILM materials were collected on a Bruker AXS D8 Advance Bragg–Brentano geometry with Cu sealed˚ plus a secondary beam tube radiation source (l ¼ 1:54178 A) graphite monochromator. A step of 0:02 and a time of 3 s step1 were selected.

2.3.

Sorption experiments

2.3.1.

Cesium sorption studies

The uptake of Csþ was investigated via equilibration and exchange kinetic measurements. The conventional batch technique was employed for the equilibration measurements in which a known amount of material was contacted with a solution containing chloride salts of Csþ , traced with 137Cs. After attaining equilibrium, the two phases were separated by centrifugation and the g activity of the supernatant was measured with a 22% HPGe detector. In addition to batch kinetic measurements, a dynamic dialysis method was also used (Karamanis et al., 1997). Periodically removing the dialysis bag and measuring the remaining solution activity determined the rate of cation uptake.

ARTICLE IN PRESS WAT E R R E S E A R C H

2.3.2.

Copper sorption studies

The aqueous solutions of Cu2þ were obtained from dilutions of stock solutions prepared from dissolving CuCl2. All sorption experiments were conducted in mechanically stirred glass reactors equipped with a thermometer and a pH electrode to measure variations in temperature and pH. At equilibrium, PILMs were filtered through a 0:47 mm millipore filter. Sorption kinetics were studied with the batch or the dialysis method. In each experiment, duplicate samples of 1 mL from the sorption solution were analyzed by XRF prior to materials immersion in the solution and till equilibrium attainment. The volume of 1 mL of solution was pipetted onto a 12 mm Whatman No. 42 filter (held in a mylar film of 6:3 mm thick) under continuous drying with an IR-lamp and was directly XRF measured. The quantity of metal ion sorbed on each material was determined by the difference between the initial metal concentration and the remaining concentration at equilibrium. XRF was calibrated with standards, which were prepared by diluting known volumes of the standard stock solutions in deionized water. Metal concentrations of 1.0–100:0 mg L1 were used and four replicates were prepared of each. The calibration curve was linear in all the studied region and with a minimum detection limit of 0:09 mg of copper on filter.

3.

Results and discussion

3.1.

PILMs characterization

1899

4 1 (200 7) 189 7 – 190 6

Elemental analysis of NaM and PILMs is shown in Table 1. The increase in aluminum weight percentage between NaM and the PILM derivatives was dependent on the initial aluminum availability per gram of clay. It is known that the Al/clay ratio has a significant effect on the accessibility properties in Alpillared clays. With an Al/clay ratio of higher than 5 mmol g1 , a bimodal micropore distribution has been observed with peak dimensions at 0.4 and 0.55 nm (Hutson et al., 1999). Moreover, higher initial aluminum availability leads to a higher pillar density (Gil and Montes, 1995) and therefore higher fraction of pores with an opening of less

than 0.45 nm. In this way, the effect of the pillar density or the interpillar distance on cesium or copper sorption was further studied. The grain size of all the prepared PILMs was measured to be lower than 45 mm. The variation of the OH/Al ratio resulted in different d001 -spacing of the pillared samples as deduced from the PILMs XRD patterns (Fig. 1). The d001 peak was narrower and more intense in Al2MFD15 (1.75 nm) and Al2MFD5 (1.74 nm) than in the other two pillared samples Al2MFD3 (1.72 nm) and Al1MFD3 (1.67), indicating a more ordered or higher crystallinity in these samples. The OH/Al ratio of 0.125 did not result in a pillared sample and thus, the sample was not used in the subsequent sorption tests. All other samples were tested in sorption experiments and the influence of PILMs interlayer spacing in their sorption ability was further investigated.

Fig. 1 – XRD patterns of sodium montmorillonite (NaM) and the prepared pillared samples (01MFD3 is the sample with OH/Al ratio of 0.125 and Al/clay ratio of 3 mmol g1 , 1MFD3 is Al1MFD3 with an OH/Al ratio of 1 and Al/clay ratio of 3 mmol g1 , etc.).

Table 1 – Elemental analysis of Na-montmorillonite (NaM) and its aluminum pillared products (%) SiO2c Al2O3c Fe2O3d MgOd CaOc K2Od Na2Oc a b c d

NaM

Al1aMFD3b

Al2aMFD3b

Al2aMFD5b

Al2aMFD15b

55.9 18.6 2.9 2.9 0.1 0.5 2.5

52.2 24.3 2.9 2.7 0.1 1.0 2.2

52.7 24.5 2.9 2.6 0.1 1.3 2.2

52.3 25.3 2.9 2.6 0.1 1.1 2.2

52.4 28.5 2.9 2.4 0.1 1.2 2.4

Molar OH/Al ratio. Al/clay ratio (mmol g1 Þ. PIGE. XRF.

ARTICLE IN PRESS 1900

WAT E R R E S E A R C H

3.2.

Cesium sorption experiments

3.2.1.

Cesium sorption on different PILMs

41 (2007) 1897– 1906

Even in the case of a nuclear accident, the concentration of cesium ions in the surface waters is extremely low (less than 1 ppb). For example, 1 MBq L1 137Cs is equivalent to 6:8  109 M. Therefore, the prepared PILMs were tested for their ability to sorb cesium in a solution of natural mineral water (0.2 L) with cationic concentrations of Ca: 1  103 M, Mg: 0:65  104 M, Na: 0:75  104 M, K: 0:2  104 M and a small added radiocesium concentration (1:7  1011 M). In these experiments, the cesium removal for a contact time of 1.5 h was measured to be around 75  2% for all the four pillared samples (20 mg). Therefore, either the d001 interlayer spacing or the interpillar distance had no effect in cesium sorption from dilute concentrations.

3.2.2.

Fig. 3 – Effect of pH on cesium sorption (1:7  1011 M) on Al2MFD15 (0:15 g L1 ).

Cesium sorption kinetics

Different kinetic models such as the Lagergen’s pseudo-firstorder equation, second-order equation and Elovich equation were tested to find out which model is in agreement with the experimental results of the kinetic study. On comparison, the pseudo-first-order rate equation yielded the best results for cesium sorption on PILMs. It can be represented as dq ¼ k1  ðqe  qÞ. dt

(1)

Integrating Eq. (1) for the boundary condition qt ¼ 0 at t ¼ 0, the equation becomes qt ¼ qe  ð1  ek1 t Þ.

(2) 1

The rate constant k1 (min Þ and the equilibrium capacity, qe ðmmol g1 Þ were determined as free parameters by a ‘‘bestfit’’ minimization of the corresponding w2 function. The kinetic isotherms and their corresponding fitting results in Fig. 2 showed that Csþ sorption in PILMs is a fast process and is accomplished in less than 30 min. The rate constant was the same for the different PILMs while the equilibrium capacity was higher for the Al2MFD15 material due to a higher restoration of the cation-exchange capacity of the initial NaM material. A similar behavior was also observed in kinetics conducted in the presence of the competing cations of Kþ or Ca2þ.

3.2.3.

Effect of pH

The influence of pH on the sorption of cesium was examined in the pH range from 3.0 to 8.0. As shown in Fig. 3, the percentage of sorbed cesium increased with increasing pH with a plateau region from around 4.5 to 8.0. The decrease in the Cs uptake observed at low pH values can be partly attributed to the positive charge, which develops on the PILM edges according to the process noted by Avena et al. (1990) SO þ Hþ ! SOH and SOH þ Hþ ! SOHþ2 where S stands for any surface site. However, it has previously been reported (Delgado et al., 1986) that the clay sheets of montmorillonites remain negatively charged at pH values down to 3, since the positive charge of the edges does not exceed the charge of the sheets. It can therefore be concluded that the Cs uptake at acidic pH values is suppressed, because the remaining negative charge of the PILM is preferentially compensated by H3 Oþ ions. Hydronium ions, although monovalent, behave mostly as di or trivalent ion. The same behavior is also usually observed in non-pillared exchanging clays (Grim, 1968). In contrast, the observed increase of cesium sorption with increasing pH can be attributed to the decrease of the competition of the hydronium ions for PILMs’ sites at higher pH values.

3.2.4.

Cesium sorption isotherms

Equilibrium sorption studies were performed to determine the maximum cesium sorption capacity of PILMs in the availability range of 0.4–10 mmol cesium per gram of PILMs in 200 mL aqueous solution. Cesium uptake was quantitatively evaluated using the one or two-site Langmuir model: qe ¼

m X i¼1

Fig. 2 – Cesium (1 mM) sorption kinetics by Al1MFD3 and Al2MFD15 (0:5 g L1 ).

Ki  Ce Q, 1 þ Ki  Ce i

(3)

where qe is the adsorption capacity at equilibrium (mg g1 Þ, Ce is the equilibrium concentration of metal ions in the solution (mg L1 Þ, m is the number of energetically different sorption sites, Q i and Ki are the adsorption capacities and binding strengths of the adsorbed cations, respectively.

ARTICLE IN PRESS WAT E R R E S E A R C H

4 1 (200 7) 189 7 – 190 6

1901

Table 2 – Freundlich and Langmuir isotherm values and parameters for the sorption of Csþ onto Al2MFD15 sample Values and parameters

Magnitude

Freundlich 1:40  0:08

KF (L g1 Þ 1=n

0:23  0:04 40.53

w2 One– site Langmuir Q ðmg g1 Þ

82:64  2:05

K ðL mg1 Þ

0:10  0:01

w2 SD(%)

5.381 5.68

Two-site Langmuir 18:56  26:30

Q 1 ðmg g1 Þ K1 ðL mg1 Þ

0:74  0:85

Q 2 ðmg g1 Þ

66:12  25:32

K2 ðL P mg1 Þ

0:066  0:034

w2 SD(%)

4.95

low sorption maximum (20%) and low-affinity sites exist with a high sorption maximum (80%). However, the error in the estimated parameters was high and the exact amount of different sites could not be safely concluded. Therefore, the surface heterogeneity of PILMs was further investigated by varying the initial cesium concentration in the presence of competitive cations.

2.52

The Freundlich model was also used as log qe ¼ log KF þ

1 log Ce , n

Fig. 4 – Comparison of the Langmuir and Freundlich isotherms for the sorption of cesium onto NaM and Al2MFD15 (qe is the adsorption capacity at equilibrium (mg g1 ) and Ce is the equilibrium concentration of metal in the solution (mg L1 )).

(4)

where qe and Ce as described in the Langmuir model, b and n are constants related to the energy of adsorption and KF is a constant related to sorption capacity. The values of normalized standard deviation (SD(%)) were calculated using the equation sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi P exp exp 2 ½ðqe  qcal e Þ=qe  , (5) SDð%Þ ¼ 100  N1 where the superscripts ‘‘exp’’ and ‘‘cal’’ are the experimental and calculated values with the fitted parameters and N is the number of measurements. The values of the isotherm constants in Table 2 were determined by fitting the above equations to the experimental data. To assess the different isotherms and their validity to correlate experimental results, the experimental data and the theoretical plots for each isotherm are shown in Fig. 4. The one-site Langmuir represented better the cesium sorption data than the Freundlich isotherm (Table 2), suggesting the monolayer sorption, mainly due to ion exchange. The w2 and (SD(%)) values for the one-site Langmuir model were very low, indicating a very good mathematical fit. The cesium sorption capacity of PILMs was determined to be less than that of the starting clay and this can be attributed to the remaining charge of aluminum pillars and/or the incomplete cationexchange capacity restoration. As it was observed in kinetic isotherms, the maximum cesium sorption capacity of the PILM materials was obtained with sample Al2MFD15. Moreover, the two-site Langmuir model fitted the experimental results even better, indicating the existence of heterogeneity. The two-site Langmuir constants for the sorption of Csþ showed that high-affinity sites exist with a

3.2.5.

PILMs selectivity for cesium sorption

Cesium selectivity of the PILM materials was studied for the binary systems of Csþ =Naþ , Csþ =Kþ and Csþ =Ca2þ in several ionic concentrations from dilute up to highly concentrated solutions of cesium and competitive cations. In all systems, materials were initially transformed in the form of the competing cation. Assuming the general exchange reaction zst;A A þ zst;b B 2 zst;A A þ zst;B B

(6)

the cesium selectivity was studied through: A. Plots of the equivalent fractions of cesium sorbed in PILMs (xCs Þ vs. their equivalent fraction in the solution (xCs Þ at equilibrium for constant solution ionic strength (IN Þ. B. Variation of the distribution coefficient Kd defined as the ratio of concentrations of Cs sorbed xc , (mol kg1 Þ and in the solution [Cs] (mol L1 Þ. C. Variation of the selectivity coefficient Kc defined as Kc ¼

ðY A Þzst;A ðMB Þzst;B ðgB Þzst;B   , z z ðY B Þzst;B ðMA Þ st;A ðgA Þ st;A

(7)

where Y A are the equivalent fraction of ions in the solid, MA the molality of ions in the solution and gA the activity coefficients in the liquid phase (Ioannidis et al., 2000). These were calculated with the extended Debye-Hu¨ckel equation. Starting from the first representation, the results in Fig. 5 clearly demonstrate that PILMs exhibit a high selectivity for Csþ sorption over Kþ and much higher over Naþ. By varying the initial concentration of Cs in the solution (108 2104 MÞ, a

ARTICLE IN PRESS 1902

WAT E R R E S E A R C H

41 (2007) 1897– 1906

decrease of the distribution coefficient Kd was observed and shown in Fig. 6. The variation of Kd with the initial Cs concentration shows that the adsorption isotherm is not linear (Staunton and Roubaud, 1997). The decrease of the Kd coefficient was much higher in the system Cs–Na-PILM than the Cs–K-PILM and this order was observed at all levels of Csþ , Naþ or Kþ concentrations. In the same figure, the variation of the distribution coefficient for two constant cesium concentrations as a function of the concentration of the competitive Naþ or Kþ (2  104 20:5  101 M) is also included. The slope of the Kd reduction was greater for K-PILM than Na-PILM for the two constant cesium concentrations. However, the Kd parameter is not a sensitive measure of the relative sensitivities of a material for various cations since the activity coefficients and the stoichiometry of the sorption process are not included in the Kd definition (Staunton and Roubaud, 1997). To overcome this weakness, the variation of the selectivity coefficient Kc as a function of the sorbed cesium fraction (zCs) in PILMs was studied. As shown in Fig. 7 for the

Fig. 5 – Cs–Na and Cs–K selectivity isotherms on PILMs in aqueous solutions of total ionic strength of 103 .

sodium form of pillared clays, values of the selectivity coefficient were high for 1–2% of site occupancy and started to decrease with increasing Cs concentration and higher sorption. Similar results were also observed with potassium as the interlayer exchangeable cation. These results indicate that energetically less favorable sites become involved with increasing cesium concentration in the aqueous solution. Such behavior is typical for heterogeneous materials with sites of different selectivity (Zachara et al., 2002). In PILMs, the selectivity values for this small fraction of sites are much higher than those of the starting NaM material and comparable to those found in materials with high cesium selectivity as ferrocyanides (Lin et al., 2001), a small fraction of the total sites of illite (Staunton and Roubaud, 1997; Zachara et al., 2002) and titanosilicates (Solbra et al., 2001). The fraction of the highly selective sites of PILMs for cesium sorption can be attributed either to micropores with a suitable opening for cesium uptake (around 0.4 nm according to Hutson et al., 1998) or to collapse of the aluminosilicate sheets and creation of sites similar to illite.

Fig. 7 – Variation of the selectivity coefficient Kc as a function of the cesium fraction in PILMs for sodium solution of 1  103 M.

Fig. 6 – Variation of the distribution coefficient Kd as a function of cesium, sodium or potassium concentration.

ARTICLE IN PRESS WAT E R R E S E A R C H

1903

4 1 (200 7) 189 7 – 190 6

In order to clarify the origin of PILMs heterogeneity, different blocking agents were used to isolate the PILMs’ group sites. Lithium ions were used to block any sites associated with the clay’s hexagonal cavities. Silver thiourea ions (AgTUþ Þ, formed from a mixture of SCðNH2 Þ2 and AgNO3 in a molar ratio of 3:1, were used to mask the regular interlamellar (planar) sites (Cremers et al., 1988). Moreover, it has been observed that the crystal violet (CV) cation is strongly attached at the surface of PILMs since its diffusion is hindered in PILMs interlammelar space due to steric reasons (Mishael et al., 1999). Therefore, CV was used to block the surface and edge sites. Prior to the experiments with blocking agents all pillared samples were saturated and transformed in the form of the blocking agent in concentrations of at least 10-fold of PILMs cesium sorption capacity. Cesium sorption isotherms were measured by equilibrating the agent-saturated PILMs with cesium solutions of varying initial concentration in a background ion concentration of 1  103 M. As seen in Fig. 8, the values of the selectivity coefficient Cs–Na for the NaM montmorillonite are in agreement with those usually observed (Staunton and Roubaud, 1997). The values of the selectivity coefficient Cs–Na for the PILMs material are much higher than those of the starting clay in all the cesium concentration range. The values of the selectivity coefficient Cs–Li follow the same behavior as those of the Cs–Na selectivity and are almost the same. Therefore, the high-affinity sites are not related with the hexagonal

Fig. 8 – Variation of the selectivity coefficient Kc as a function of the cesium fraction in Al1MFD3 for different blocking agents of 1  103 M.

cavities of the clay structure. The values of the selectivity coefficient Cs–Ca for 1–2% of site occupancy are comparable to the very high values observed in a 0.1% fraction of illite sites. The silver thiourea cation does not affect the high affinity sites for cesium sorption but reduces the low affinity sites in the same manner as in illite (Zachara et al., 2002). The CV cation reduces all sites by strongly limiting the highaffinity sites but also blocking (due to the formation of molecular aggregates on the outer surface of the clay (Mishael et al., 1999)) the pores entrance and thus reducing the lowaffinity sites as well. Concluding, the results of the present study provide direct evidence that the high selectivity of PILMs for cesium sorption is rather related to their surface and edge sites than to micropores of less than 0.4 nm opening. This conclusion is corroborated by the fact that if the selectivity was due to micropores, the high-affinity sites should account for more than 15% of the total sites and in accordance to the ratio of the micropore volume to the total volume (Hutson et al., 1998). Moreover, an increase in the micropore volume of less than 0.4 nm opening has been observed with increasing ionic radius of the post exchange cation (e.g. cesium) (Hutson et al., 1998). In this case, the cesium loading should lead to higher amount of pores with opening less than 0.4 nm and subsequently higher selectivity. This result was not observed in the present study. Finally, increasing the pillaring density and therefore the fraction of micropores, did not result in higher selectivity coefficients.

3.3.

Copper sorption experiments

3.3.1.

Copper sorption on different PILMs

Considering the 0.96 nm layer thickness of montmorillonite and the 0.54 nm diameter of the hydrated copper cation CuðH2 OÞ2þ (Kukkadapu and Kevan, 1988), a minimum d001 6 spacing of 1.50 nm is necessary for the existence of the species in PILMs interlayer space. Thus, the CuðH2 OÞ2þ 6 hydrated copper ion can enter the interlammelar space of all the prepared aluminum-pillared montmorillonites. In this frame, sorption of copper (0.5 mM) on the prepared PILM materials was studied in aqueous solutions without or with the presence of competing cations of Hþ , Kþ or Ca2þ in concentrations similar to that found in drinking water (1.5 mM). As seen in Table 3, the Al2MFD15 sample exhibited the higher sorption ability of all the prepared materials and its

Table 3 – Copper sorption from aqueous solutions by different PILMs (total concentration ¼ 0:5 mM)

Cu sorbed (%) Cu sorbed (meq g1 )/Cs sorption capacityb (meq g1 Þ

NaM

Al1MFD3

Al2MFD5

Al2MFD15-Una

Al2MFD15

60:5  2:3 0.75

42:8  1:8 1.47

67  2:5 1.49

20:7  1:6

95:1  2:9 1.61

The solution/solid ratio ðV=MÞ ¼ 1 L g1 . pH ¼ 4:8. Al2MFD15-Un is the sample that the cation-exchange capacity was not restored after pillaring. b Cs sorption capacity is assumed to be equal to the cation-exchange capacity (CEC) of the sample. a

ARTICLE IN PRESS 1904

WAT E R R E S E A R C H

41 (2007) 1897– 1906

sorption capacity was higher than the cation-exchange capacity of the starting clay material. Assuming the equality of the maximum cesium sorption capacity and the cationexchange capacity of the PILMs materials (Karamanis et al., 1997), the second row in Table 3 indicates that the amount of Cu2þ sorbed on PILMs increased with higher pillar density. Therefore, copper sorption on PILMs involves both a cationexchange process in the surface and interlammelar clay sites and complexation reactions with the pillar oxides. This conclusion is further corroborated from the observation that the PILM sample with unrestored cation-exchange capacity, sorbed 0.21 meq of copper per gram of material. This amount is mainly bound to pillar oxides since PILMs’ cation-exchange capacity is drastically reduced after the pillaring process (Karamanis et al., 1997). Furthermore, a reduction of 21.3% and 15.3% of sorbed copper by Al2MFD15 was observed with the addition of potassium or calcium ions as competitive cations. Both reductions can be attributed to the reduction of the sorption fraction of the interlayer sites group due to PILMs preference for the sorption of monovalent cations as potassium over divalent as calcium. This result was also observed in cesium sorption. Finally, the Al2MFD15 sample with a solution to solid ratio (V=M) of 1 L g1 , removed 99.7% of an initial copper concentration of 32 mg L1 in natural mineral water, reducing in this way the copper level in a value much lower than the maximum acceptable level of 2 mg L1 .

3.3.2.

Copper sorption kinetics

Copper sorption on the PILM material with the highest Al/clay ratio (Al2MFD15) was fast and was accomplished within 20 min (Fig. 9). After testing different models, the copper sorption by PILMs was better described by the Lagergren pseudo-first-order equation. The rate constant k1 (min1 ) and the equilibrium capacity, qe ðmmol g1 Þ determined by fitting Eq. (2) to the experimental results are included in Fig. 9. The rate constant of copper sorption on the pillared sample was higher than the starting montmorillonite. This result can be attributed to the easiness of pores accessibility due to the three-dimensional structure of PILMs than the blocking of pores after the initial sorption of copper within the montmorillonite interlayer space and the subsequent collapse of the aluminosilicate clay sheets. Concluding, it is clear from the kinetic measurements that the velocity of transport of copper from the liquid phase to solid phase is rapid enough for PILMs application purposes in the treatment of polluted aqueous solutions.

3.3.3.

Fig. 9 – Copper (0.5 mM) sorption kinetics by NaM and Al2MFD15.

Effect of pH

The effect of the pH on the copper sorption by PILMs was studied in the pH region between 3.0 and 6.0. The pH was limited to values less than 6 because of the formation of copper hydroxyl species at higher pH as determined by the visual MINTEQ code (Gustafsson, 2003). As seen in Fig. 10, the sorption of copper ions increased around 30% with an increase in pH of the solution from 3.0 to 4.0 and then remains constant. As it has been observed in similar studies with PILMs, the increase of pH decreases the competition between the protons and the metal ions for surface sites and results in increased metal uptake by the PILM (Manohar et al.,

Fig. 10 – Effect of pH on copper sorption (0.5 mM) on NaM and Al2MFD15.

2006). Finally, copper sorption on the Al2MFD15 sample was much higher than the removal by montmorillonite in all the studied pH range.

3.3.4.

Copper sorption isotherms

The equilibrium data for copper sorption covered the concentration range from 10 to 100 mg L1 and were subjected to the sorption isotherms of Langmuir and Freundlich (Fig. 11). The Langmuir model parameters and the statistical fits of the sorption data to Eqs. (3) and (4) are given in Table 4. The one-site Langmuir isotherm appeared to be inadequate to describe the sorption behavior of PILMs because it underestimated sorption at low initial concentrations. The high SD(%) value of the one-site Langmuir model was dramatically reduced with the application of the Freundlich model. The Freundlich model is characterized by 1=n, the heterogeneous factor, hence it is applicable to sorption on heterogeneous surfaces, i.e., surface with non-energetically equivalent sites. The fact that the Freundlich isotherm fits the experimental data very well can be explained from the heterogeneous distribution of active sites on the PILM materials. Indeed, the two-site Langmuir model described copper sorption onto

ARTICLE IN PRESS WAT E R R E S E A R C H

4 1 (200 7) 189 7 – 190 6

4.

Fig. 11 – Comparison of the Langmuir and Freundlich isotherms for the sorption of copper onto NaM and Al2MFD15 (qe is the adsorption capacity at equilibrium (mg g1 ) and Ce is the equilibrium concentration of metal in the solution (mg L1 )).

Table 4 – Freundlich and Langmuir isotherm constants for the sorption of Cu2þ onto Al2MFD15 Values and parameters

Magnitude

Freundlich KF (L g1 Þ 1=n R2 SD(%)

1905

Conclusion

The results of this study indicate that Al-pillared montmorillonites are potential sorbents for the removal of cesium or copper from aqueous solutions. The most effective pH range was found to be 4.0–6.0 for the removal of copper and 3.0–8.0 for cesium. The sorption of either cesium or copper follows the pseudo-first-order sorption reaction while the sorption isotherms follow the two-site Langmuir model. Sorption experiments with blocking agents revealed that complementary to the interlayer clay sites, a small fraction of PILMs sites exist (1–2%) that is very selective for cesium sorption over competitive monovalent and divalent inorganic and organic cations. These sites are related to the surface or edges of the PILM materials and their fraction is independent on PILMs d001 spacing or the Al/clay ratio used in their preparation. In contrast, the Al/clay ratio appears to have a significant effect on the sorption of copper on PILMs. With increasing Al/clay ratio, the amount of sorbed copper increases. This indicates that the sorption of copper involves specific group of highaffinity sites on the pillar surfaces in addition to the lowaffinity restored clay interlayer sites of the PILM materials. Therefore, copper sorption should be driven by both a cationexchange mechanism and by complexation reactions with the pillar oxides.

19:53  1:02 0:112  0:006 0.995 2.78

Acknowledgments

One-site Langmuir Q ðmg g1 Þ

32:92  1:41

K ðL mg1 Þ

0:394  0:275

R2 SD(%)

0.998 43.84

Two-site Langmuir

R E F E R E N C E S

Q 1 ðmg g1 Þ

23:89  4:36

K1 ðL mg1 Þ

0:011  0:004

Q 2 ðmg g1 Þ

22:12  0:50

K2 ðL mg1 Þ

15:97  1:69

w2 SD(%)

This work was partially supported by the Empirikion Foundation. The authors thank Dr. N. Kourkoumelis of the XRD Unit of the University of Ioannina.

0.07 0.45

PILMs even better. The determined constants showed that high and low affinity sites exist with almost the same capacity and with a value half to the sorption maximum (0:72 mmol g1 Þ. These two different affinity sites can be attributed to the pillar oxides and to typical exchange sites compensating the negative charge of the clay sheets. The last are restored after the pillaring process. In desorption studies, the Cu2þ ions were detected in negligible quantities for all samples when the desorption was carried out using deionized water. When the supernatant was replaced by CaCl2 or KCl solutions, very small amounts of Cu2þ were detected in the solutions. In contrast, by lowering the pH from about 5–6 after the end of sorption to around 1.8, almost all the sorbed copper was released in the solution.

Al-Qunaibit, M.H., Mekhemer, W.K., Zaghloul, A.A., 2005. The adsorption of Cu(II) ions on bentonite—a kinetic study. J. Colloid Interface Sci. 283, 316–321. Apak, R., Atun, G., Gu¨c-lu¨, K., Tu¨tem, E., 1996. Sorptive removal of cesium-137 and strontium 90 from water by unconventional sorbents. II. Usage of coal fly ash. J. Nucl. Sci. Technol. 33, 396–402. Avena, M.J., Cabrol, R., De Paulil, C.P., 1990. Study of some physicochemical properties of pillared montmorillonites: acid–base potentiometric titrations and electrophoretic measurements. Clays Clay Miner. 38, 356–362. Babel, S., Kurniawan, T.A., 2003. Low-cost adsorbents for heavy metals uptake from contaminated water: a review. J. Hazard. Mater. 45, 219–243. Bailey, S.E., Olin, T.J., Bricka, R.M., Adrian, D.D., 1999. A review of potentially low-cost sorbents for heavy metals. Water Res. 33, 2469–2479. Cooper, C., Jiang, J.Q., Ouki, S., 2002. Preliminary evaluation of polymeric Fe- and Al-modified clays as adsorbents for heavy metal removal in water treatment. J. Chem. Technol. Biotechnol. 77, 546–551. Cremers, A., Elsen, A., De Preter, P., Maes, A., 1988. Quantitative analysis of radiocesium retention in soil. Nature 335, 247–249. Delgado, A., Gonzalez-Caballero, F., Bruque, J.M., 1986. On the zeta potential and surface charge density of montmorillonite in

ARTICLE IN PRESS 1906

WAT E R R E S E A R C H

41 (2007) 1897– 1906

aqueous electrolyte solutions. J. Colloid. Interface Sci. 113, 203–211. Gil, A., Montes, M., 1995. Analysis of the microporosity in pillared clays. Langmuir 10, 291–297. Grim, R.E., 1968. Clay Mineralogy. McGraw-Hill, New York, p. 214 (Chapter 7). Gurboga, G., Tel, H., 2005. Preparation of TiO2–SiO2 mixed gel spheres for strontium adsorption. J. Hazard. Mater. 120, 135–142. Gustafsson, J.P., 2003. Visual MINTEQ (VMINTEQ) Version 2.14; KTH, Department of Land and Water Resources Engineering, Stockholm, Sweden. Hutson, N.D., Gualdoni, D.J., Yang, R.T., 1998. Synthesis and characterization of the microporosity of ion-exchanged Al2O3-pillared clays. Chem. Mater. 10, 3707–3715. Hutson, N.D., Hoekstra, M.J., Yang, R.T., 1999. Control of microporosity of Al2O3-pillared clays: effect of pH, calcination temperature and clay cation exchange capacity. Microp. Mesop. Mater. 28, 447–460. Ioannidis, S., Anderko, A., Sanders, S.J., 2000. Internally consistent representation of binary ion exchange equilibria. Chem. Eng. Sci. 55, 2687–2698. Jiang, J.Q., Zeng, Z., 2003. Comparison of modified montmorillonite adsorbents. Part II: the effects of the type of raw clays and modification conditions on the surface properties and adsorption performance of modified clays. Chemosphere 53, 53–62. Karamanis, D., Aslanoglou, X., Assimakopoulos, P.A., Gangas, N.H., 2001. PIGE and XRF analysis of a nano-composite pillared layered clay material for nuclear waste applications. Nucl. Instrum. Methods B 181, 616–621. Karamanis, D.T., Aslanoglou, X., Assimakopoulos, P.A., Gangas, N.H., Pakou, A., Papayanakos, N., 1997. An aluminum pillared montmorillonite with fast uptake of cesium and strontium from aqueous solutions. Clays Clay Miner. 45, 709–717. Kukkadapu, R.K., Kevan, L., 1988. Synthesis and electron spin resonance studies of copper-doped alumina-pillared montmorillonite clay. J. Phys. Chem. 92, 6073–6078. Lahiri, S., Roy, K., Bhattacharya, S., Maji, S., Basu, S., 2005. Separation of 134Cs and 152Eu using inorganic ion exchangers, zirconium vanadate and ceric vanadate. Appl. Radiat. Isot. 63, 293.

Lin, Y., Fryxell, G.E., Wu, H., Engelhard, M., 2001. Selective sorption of cesium using self-assembled monolayers on mesoporous supports (SAMMS). Environ. Sci. Technol. 35, 3962–3966. Lothenbach, B., Furrer, G., Schulin, R., 1997. Immobilization of heavy metals by polynuclear aluminium and montmorillonite compounds. Environ. Sci. Technol. 31, 1452–1462. Manohar, D.M., Noeline, B.F., Anirudhan, T.S., 2005. Removal of vanadium(IV) from aqueous solutions by adsorption process with aluminum-pillared bentonite. Industrial & Engineering Chemistry Research 44, 6676–6684. Manohar, D.M., Noeline, B.F., Anirudhan, T.S., 2006. Adsorption performance of Al-pillared bentonite clay for the removal of cobalt(II) from aqueous phase. Appl. Clay Sci. 31, 194–206. Mishael, Y.G., Rytwo, G., Nir, S., Crespin, M., Bergaya, F.A., Van Damme, H., 1999. Interactions of monovalent organic cations with pillared clays. J. Colloid Interface Sci. 209, 123–128. Pendelyuk, O.I., Lisnycha, T.V., Strelko, V.V., Kirillov, S.A., 2005. Amorphous MnO2–TiO2 composites as sorbents for Sr2þ and UO2þ 2 . Adsorption 11, 799–804. Reddad, Z., Gerente, C., Andres, Y., Le Cloirec, P., 2002. Adsorption of several metal ions onto a low cost biosorbent: kinetic and equilibrium studies. Environ. Sci. Technol. 36, 2067–2073. Solbra, S., Allison, N., White, S., Mikhalovsky, S., Bortun, A.I., Bortun, L.N., Clearfield, A., 2001. Cesium and strontium ion exchange on the framework titanium silicate M2 Ti2 O3 SiO4  nH2 O (M ¼ H, Na). Environ. Sci. Technol. 35, 626. Staunton, S., Roubaud, M., 1997. Adsorption of 137Cs on montmorillonite and illite: effect of charge compensating cation, ionic strength, concentration of Cs, K, and fulvic acid. Clays Clay Miner. 45, 251–260. Stout, S.A., Cho, Y., Komarneni, S., 2006. Uptake of cesium and strontium cations by potassium-depleted phlogopite. Appl. Clay Sci. 31, 306–313. Varma, R.S., 2002. Clay and clay-supported reagents in organic synthesis. Tetrahedron 58, 1235–1255. Wang, Y.H., Lin, S.H., Juang, R.S., 2003. Removal of heavy metal ions from aqueous solutions using various low-cost adsorbents. J. Hazard. Mater. B 102, 291–302. Zachara, J.M., Smith, S.C., Liu, C., McKinley, J.P., Serne, R.J., Gassman, P.L., 2002. Sorption of Csþ to micaceous subsurface sediments from the Hanford site, USA. Geochim. Cosmochim. Acta 66, 193–211.