Water treatment residuals source and preparation
Water treatment residuals (WTRs) from two water treatment plants in Staffordshire, England, were provided by Severn Trent Water. One plant primarily uses Al salts and the other Fe-based salts, generating what is designated here as Al-WTRs and Fe-WTRs, respectively. According to the results from a previous project carried out on the same materials (Howells et al. 2018), the properties of Al-WTRs once dried were pH 7.34 ± 0.06, Al content 11.64 ± 1.08% w/w, organic matter (OM) content 28.0 ± 0.1% w/w and Fe content 0.91 ± 0.08% w/w, while those of Fe-WTRs were pH 7.37 ± 0.01, Fe 17.69 ± 0.19% w/w, OM 25.9 ± 0.2% w/w and Al 0.71 ± 0.12% w/w. For this study, the as-received (i.e. moist) WTRs were dried at 30 °C until they became a stable mass and were ground to pass a 2-mm sieve.
Water treatment residuals (WTRs) adsorption capacity for lead and zinc
Following commonly employed batch exchange procedures for determining the adsorption capacity of a substance, 2.0 g WTR samples were equilibrated with 20 mL solutions with Pb or Zn concentrations of 10, 50, 100, 200, 300 or 400 mg/L (3 replicates for each WTR type at each concentration). Samples were equilibrated via end-over-end shaking (40 rpm; ~ 20 °C) for 24 h, which is generally recognised as sufficient time to establish equilibrium (e.g. Coles and Yong 2006). Equilibration was immediately followed by centrifugation, filtration of supernatant using 0.45-µm syringe filters, acidification with a drop of trace analysis grade concentrated nitric acid (Primar Plus) and then analysis for metals using ICP-OES (Varian vista) (e.g. as per Dada et al. 2012). Concentrations determined in the equilibrium solutions were used to calculate amounts of metal sorbed to WTRs, and the data were further examined through constructing Henry’s, Langmuir, Freundlich and Temkin adsorption isotherms models, as outlined in Table 1. Equation 1, which was reported by Vanderborght and Van Grieken (1977), is used to calculate the quantity of sorbate retained by a unit of mass of the sorbent Qeads (mg/g).
$$ Q_{e} = \left[ {\left( {C_{0} - C_{e}^{ads} } \right){*}V} \right]/m $$
(1)
where C0 = initial solution concentration before adsorption; Ceads = concentration at adsorption equilibrium; V = volume of the adsorbate, and m = mass of WTRs in grams.
The adsorbed amount of Zn or Pb by each of Al-WTR and Fe-WTR can be expressed as an adsorption percentage (Eq. 2), which is based on the ratio of the mass of adsorbed ions at equilibrium to the initial mass of adsorbate ions in solution (OECD 2000).
$$ {\text{Adsorption \% }} = \frac{{\left( {{\text{C}}_{0 - } {\text{C}}_{e}^{ads} } \right){\text{*V}}}}{{\left( {C_{0} {\text{*V}}} \right)}} \times 100\% $$
(2)
At its simplest, adsorption of a metal (or other substance) from solution into solid can be expressed as Henry’s adsorption isotherm, with KH typically denoted as Henry’s adsorption constant (units of L/g). However, adsorption onto a solid across a wide concentration range is not typically constant. In order to understand some of the adsorption characteristics of Zn and Pb into WTRs, the experimental data were fitted to Henry’s, Langmuir, Freundlich and Temkin linearised equations, all of which have been described in detail elsewhere (Dada et al. 2012; Sparks 2003; Yildirim 2006). From a fitted Langmuir isotherm, one can calculate the Qm value, which refers to the maximum monolayer coverage capacity (mg/g) onto the adsorbent, assuming that the binding energy of the sites is homogenous and can also determine the ‘b’ value which is the term used for the Langmuir isotherm constant (L/mg) (Dada et al. 2012; Bonilla-Petriciolet et al. 2017).
Table 1 Linear and nonlinear isotherm equations (Yildrim 2006; Dada et al. 2012) According to the Temkin isotherm, the heat of the adsorption for all adsorbates decreases linearly as the amount of the adsorbed materials on the sorbent increases. An important calculated parameter for this isotherm is often designated the ‘B’ value, which is the constant related to heat of sorption (J/mol); the AT value is the Temkin isotherm equilibrium constant (L/g) (Dada et al. 2012; Bonilla-Petriciolet et al. 2017). For adsorptions onto adsorbents for which the distribution of heat of adsorption of binding sites is not or cannot be assumed to be uniform, the Freundlich isotherm model is often used. It is a widely applied isotherm for describing adsorption characteristics and generates the following parameters and constants: 1/n, which is dimensionless and is a function of the strength of the adsorption; and Kf is the Freundlich isotherm constant (with units of mg 1−(1/n).g−1.L(1/n)) (Dada et al. 2012; Bonilla-Petriciolet et al. 2017). Although widely used for the purpose, determining characteristics of adsorption processes based on the comparison of Freundlich Kf values can be problematic when 1/n values are not the same or Ceads ≠ 1 because the units of Kf will be different (Chen et al. 1999). To address this, Chen et al. (1999) proposed a unified adsorption variable (Kuads) to unify the unit of Kfads to be L/g. The Kuads is the slope of the isotherm at any value of Ceads or Qeads and can be calculated over a range of Ceads or Qeads using Kfads and 1/n (using Eq. 3 or Eq. 4). Therefore, Kuads is also calculated in the present study:
$$ {\text{K}}_{u}^{ads} = {\text{K}}_{f}^{ads} /C_{{\text{e}}}^{{ads \left( {n - 1} \right)/n}} $$
(3)
$$ {\text{K}}_{u}^{ads} = {\text{K}}_{f}^{ads n} /Qe^{ads n - 1} $$
(4)
Thermodynamic data such as the standard adsorption energy can be obtained from Langmuir and Temkin equation using Eq. 5 (Kim et al. 2004):
$$ {\text{K}} = e^{{ - \Delta G_{ads}^{0} /RT}} $$
(5)
Desorption of Pb and Zn
The desorbability of Zn and Pb bound to Al-WTRs and Fe-WTRs was determined in batch desorption experiments to determine the degree of reversibility of adsorption by the materials, that is, to determine how well they retain the pollutant metals after exposure to clean solution. Desorption experiments were conducted on the 2 g samples after removal of the supernatant following the initial equilibration process described above for the adsorption experiment. In order to facilitate accurate desorption measurements, and specifically to allow any metals remaining entrained in the WTRs after solution removal to be fully accounted and adjusted for in the calculations, the mass of each tube had been recorded before adding Pb or Zn solution in the sorption experiment and was recorded again at the end of the batch adsorption equilibrium experiments (i.e. after centrifugation and solution removal). The difference in mass enabled calculation of the remaining volume of solution within each tube, and this, together with the measured concentrations of metals in the removed solution, allowed calculation of the amount of entrained metals in that remaining solution.
For desorption, the removed supernatant was replaced by fresh 0.001 M CaCl2 solution and the samples were shaken (40 rpm; ~ 20 °C) again for 24 h, which was immediately followed by centrifugation, filtration using 0.45-µm syringe filters and then analysis for metals using ICP-OES. Concentrations measured in the 0.001 M CaCl2 desorption supernatant solutions (Cedes), corrected for the calculated amounts of entrained metals remaining from the initial sorption solution, were used to express the desorbed amount of Zn or Pb from each of Al-WTRs and Fe-WTRs as a desorption percentage (Eq. 6; OECD 2000).
$$ {\text{desorption \% }} = \frac{{\left( {{\text{C}}_{e}^{des} {\text{*V}}} \right)}}{{({\text{C}}_{o} {\text{*V}}) - ({\text{C}}_{e}^{ ads} *V)}} \times 100\% $$
(6)
Adsorption of Zn-Pb ions in combination
In addition to separate adsorption experiments with Zn and Pb individually, adsorption experiments with both Zn and Pb present in solution were conducted to examine competitive adsorption. The experiments were carried out as described above, including the desorption assessment, but with modified initial solution concentrations of (1) 10 mg/L (both Pb and Zn) and (2) 50 mg/L (both Pb and Zn). Statistical assessments of sorption and desorption were conducted via t tests and Mood’s median tests according to data distribution (normality) types using the Real Statistics Resource Pack software (Release 7.6; www.real-statistics.com).