All the chemicals used in this study were of analytical grade and used as received. Nitrate salts used for preparation of metal stock solutions were acquired from Sigma-Aldrich. Working solutions ranging from 1 to 800 mg/l (0.02 to 4 mmol/l) were made from the stock solutions. Adjustment of pH was made with 0.1 M Sodium hydroxide (NaOH) and 0.1 M Nitric acid (HNO3). Multimetal solutions were prepared by at first adding acid in the flasks in order to avoid any precipitation and conducting adsorption experiments immediately with fresh solutions. Analysis of solutions before adsorption experiments confirmed that metals were in ionic form.

EDSG and DTSG were synthesized by reacting EDTA- or DTPA-anhydrides with aminopropyl-modified silica gels in ethanol/acetic acid solution. The synthesis of modified surfaces followed the reported procedures [21] and more detailed information of the whole synthesis process can be found in our earlier publication [12]. The characterization of the products was conducted by elemental analysis, surface area and pore size analysis and zeta potential analysis and the properties of the studied materials are presented in Table 1.

The effective parameters such as adsorbent concentration, contact time, and pH were optimized in our earlier studies. As observed earlier the maximum metal adsorption was reached at pH 3. This value is close to the isoelectric points of both EDSG and DTSG. The used dose of adsorbent was 2 g/l and contact time 24 hours.

Batch adsorption experiments were conducted at ambient temperature. The sample tubes containing metal solution and the adsorbent were agitated on a rotary shaker. After experiment the adsorbent was separated from aqueous solution using 0.45 μm polypropylene syringe filter and the residual concentration of metal in the samples were analyzed by an Inductively Coupled Plasma Optical Atomic Emission Spectrometry (ICP-OES) model iCAP 6300 (Thermo Electron Corporation, USA). The adsorption capacities (q_e) (mg/g) of modified silica surfaces are calculated as follows:

where C_i and C_e are the initial and the equilibrium concentrations of the metal ions, respectively (mg/l); while M (g) and V (l) represent the weight of the adsorbent and the volume of the solution, respectively.

The exhausted adsorbent containing the adsorbed metals was collected for regeneration. Initially the adsorbent was loaded by the metal ions by mixing 0.02 g of EDSG or DTSG with 0.02 l of 10 mg/l metals solution. After attaining the equilibrium the adsorbent was separated from the solution with centrifuge. Metal ions were eluted by mixing the adsorbent with 5 ml of 2 M HNO₃. Effluents were collected and chemically analyzed for metal determination. The regeneration efficiency (%RE) of the adsorbent was calculated using eq. (2):

where q₀ and q_r are the adsorption capacity of original and regenerated adsorbent respectively.

At first solutions spiked with 1 μg/l of metals were prepared by pipetting 0.5 ml of 1 mg/l stock solutions (diluted from 100 mg/l standard solutions) in 500 ml flasks followed by dilution with ultrapure water. This solution was pumped with Watson Marlow peristaltic pump Sci-Q 323 through the column [minicolumns purchased from BioRad (EconoColumn, 0.7 × 10 cm)] filled with adsorbent with the flow rate of 0.5 ml/min. The volume of the adsorbent was approximately 0.75 ml. Additional rinsing was made with 20 mL of deionized water. The metal elution was conducted with 5 ml of 2 M or 3 M HNO₃ and the effluent used for metal analysis. 3% NaCl and Allard ground water were also used as matrices in order to confirm the preconcentration efficiency from the various surroundings.

To theoretically evaluate the adsorption processes of different cations a total number of seven different isotherm models were used (Table 2).

Shortly the Langmuir isotherm assumes that adsorption occurs at specific homogenous sites of the adsorbent without any interactions between the adsorbates and the Freundlich isotherm is applicable for heterogeneous surfaces [23]. The Sips and Redlich-Peterson isotherms are combinations of the Langmuir and Freundlich isotherms [24], [25]. Toth model is an empirical equation, which was derived to improve Langmuir model fittings at both low and high concentration limits and it is useful in modeling heterogeneous adsorption processes [26]. Dubinin-Reduschovich isotherm is based on the potential theory and assumes a Gaussian energy distribution. Finally, Two-site Langmuir (BiLangmuir) isotherm containing four parameters assumes that the adsorbent surface has two divergent active sites with different affinities toward target compounds [27].

In order to study metal adsorption properties of EDSG and DTSG multimetal solutions with various concentrations were prepared for batch adsorption studies. Table 3 shows that EDSG and DTSG were effective for the removal of divalent metals (as observed earlier) but failed in the removal of metallic oxyanions [Arsenic (As), Chromium (Cr)]. The partial removal of chromium, however, could be due to reduction of chromium to cationic form Cr³⁺ by EDTA or DTPA. Based on these results it was suggested that adsorption occurred through the surface chelating groups, which are able to form stable complexes with cationic metal ions. Based on the results Co, Ni, Cd, and Pb were selected for the further studies.

Experimental adsorption isotherms were measured for Co, Ni, Cd, and Pb. Isotherm models were fitted by the non-linear regression using MATLAB. Theoretical calculations included the most commonly applied Langmuir and Freundlich isotherms, but also five additional models including the Sips, Redlich-Peterson, Toth, Dubinin-Radushkevich, and BiLangmuir. Notice that Table 4 is showing concentrations of metals in mmol/g instead of mg/g for the sake of the easier comparison.

Based on the results in Table 4 and Figure 1 the BiLangmuir isotherm was most applicable to represent the adsorption processes of all the metals with quite high correlation coefficients (results for DTSG not shown here). This indicated that the surface was composed of two different kinds of active sites having different affinities towards target cations. Similar observations were done in the case of Co and Ni adsorption on EDSG and DTSG in our earlier studies [12]. Later those two different adsorption sites were attributed to different pH dependent speciations of the immobilized EDTA and DTPA groups [28].

Figure 1.

Adsorption isotherms for EDSG

It was noticed that the shapes of the experimental adsorption isotherms for Pb were clearly different compared to isotherms of other metals. Also, BiLangmuir model suggested that most of the active sites for Pb binding were similar. Generally, Pb has the lowest hydration number (it is weakly hydrated) compared to other studied metals, which presumably makes it more easily attached by the surface groups. This is supported by the higher binding energy values of Pb evaluated by the Dubinin-Raduskevich model (Tables 4 and S1). On the other hand, it should be noted that based on the stability constant values (log K) of metal chelates under concern (for EDTA – Ni: 20.11, Pb: 19.71, Co: 18.16, and Cd: 18.1, for DTPA - Ni: 22.25, Cd: 21.15, Pb: 20.95 and Co: 20.95) Ni should have the highest binding energies on both EDSG and DTSG. This discrepancy has been observed earlier and is most likely related to immobilization of chelating agents and the weakest hydration tendency of Pb [29].

Generally the binding energy values and other isotherm parameters related to surface affinity towards target compounds (log K-values) were higher for DTSG than for EDSG, which is in accordance with the higher stability constants of metal DTPA-chelates compared to corresponding chelates with EDTA. Especially, Pb seemed to have the highest attraction toward DTSG surface. For further confirmation Figure 2 shows that Pb was best adsorbed also in multimetal solutions containing the same molar amount of each metal. Ni was the second best adsorbed followed by Cd and Co.

Figure 2.

Metal removal by DTSG in the mixed metal solution as a function on initial concentration

In order to evaluate the technical applicability of the adsorbents towards real samples containing variety of different metals adsorption/desorption experiments were conducted using multimetal systems. The aim was to clarify if the adsorbent is capable of simultaneous and efficient adsorption and desorption of various metal cations. Regeneration was conducted with 2 M HNO3 and the results are presented in Table 5.

The results show that nitric acid is regenerating the adsorbed metal cations from EDSG with efficiencies up to 99.5% and from DTSG with efficiencies up to 98.5% in four consecutive regeneration processes. These results also support the higher binding energies between metals and surface bound DTPA-groups than those between metals and EDTA-groups. In the case of DTSG better regeneration could have been obtained with more concentrated nitric acid.

Preconcentration of metals is highly important in order to obtain reliable analysis results of trace amounts of metals from the various solution matrices. In this study metals were preconcentrated from pure, simulated ground water, and saline solutions. The sample matrices included single metal solutions, mixed metal solutions, mixed metal solutions with 3% NaCl and mixed metal solutions in Allard ground water. All samples were spiked with 1 μg/l of metals under study. The results are collected in Table 6. Firstly, it can be seen that the determination limits even in pure water were remarkably higher than the amounts determined after preconcentration step. In addition, the present of salt increased the limit of determination notable, especially in the case of Ni. Generally, preconcentration was quite successful with recoveries up to 99.4% for both EDSG and DTSG expect for Cd in mixed metal solutions. In the case of ground water matrix especially Ni and Cd were effectively preconcentrated and analyzed.

To explore the preconcentration properties of functionalized silica gels further, DTGS was used to preconcentrate eight different metals from a simulated seawater sample. Besides Co, Ni, Cd and Pb, preconcentration was successful also in the case of Europium (Eu), Iron (Fe), and Manganese (Mn). Especially, also rare earth elements (Eu) could be analyzed more precisely after preconcentration step with functionalized silica gels. On the other hand, almost half of the titanium was not concentrated, which might be attributed its very high stability constant with DTPA (log K = 23.05) [30].