Sediment analysis

The specific surface areas of the sediments were determined with the N2

adsorption/desorption BET method (DIN ISO 9277, 2003) by using the Autosorb-1-C (Quantachrome) surface analyzer. Sediment pH values were measured according to the international norm DIN ISO 10390 (2003) with ultrapure water and a 0.01 M CaCl2

solution. The determination of the total organic carbon (TOC) content was carried out by dry combustion of the previously with 4 M HCl treated sediment at 900 °C in a total organic carbon analyzer (Dohrmann Boat Sampler 184 S / Rosemount Dohrmann DC 70). The potential CEC (at pH = 8.1) was measured in accordance with DIN ISO 13536 (1997). The mineralogical constitutions were characterized by macroscopic observation and combined X-ray diffractometry-thermoanalysis (Siemens diffractometer D5000, Netzsch STA 409 PG Luxx).

3.3 Results and discussion

The obtained sorption coefficients of all experiments can be found in Table 3.4. The modeled parameters and column conditions of the tracer tests as well as for the experiments are listed in Table B1 and B2 in the Appendix B.

3.3.1 Experiments A: Different concentrations of Ca²⁺

A tenfold increase of the Ca²⁺ concentration resulted for sediment S1 and S2 in a reduction close to one half of the retention times for atenolol (Fig. 3.1 and Fig. 3.2).

Comparing experiment A1 with A2, Kd decreased for sediment S1 from 3.5 L kg⁻¹ (R = 18.5) to 1.3 L kg⁻¹ (R = 7.4). The same behavior was observed for sediment S2, as Kd decreased from 3.2 L kg⁻¹ (R = 22.4) in experiment A3 to 1.7 L kg⁻¹ (R = 12.1) in experiment A4. Apparently, Atenolol was biodegraded in both experiments (around 15–30% of c0).

Fig. 3.1 Experimental and modeled breakthrough curves (including tracer tests) of atenolol for sediment S1 with and without CaCl2 addition (Ca²⁺ concentrations of 40 and 400 mg L⁻¹).

Concentrations c of the breakthrough curves were normalized with the initial concentration c0. Pore volumes were calculated by normalizing the experiment duration

Increased competition with inorganic cations leads to a weaker atenolol sorption onto the sediment. The effect can solely be explained by the existence of cation exchange processes, which were already expected from Yamamoto et al. (2009), Ramil et al. (2010), and Schaffer et al. (2012). Higher Ca²⁺ concentrations lead to a shift of the exchange equilibrium in favor of the double charged Ca²⁺ and as a consequence the probability for atenolol sorption on exchange sites decreases.

According to Eq. (3.3–3.5), the contributions of hydrophobic partitioning to Kd are estimated to be <0.0030 L kg⁻¹ for sediment S1 and <0.0015 L kg⁻¹ for sediment S2.

Thus, ionic interactions are clearly dominating even at high concentrations of competing inorganic ions.

Fig. 3.2 Experimental and modeled breakthrough curves (including tracer tests) of atenolol for sediment S2 with and without CaCl2 addition (Ca²⁺ concentrations of 40 and 400 mg L⁻¹).

Concentrations c of the breakthrough curves were normalized with the initial concentration c0. Pore volumes were calculated by normalizing the experiment duration to the ideal breakthrough time of the Cl⁻ tracer.

Despite differences in the sediment properties, the sorption behavior is very similar for both sediments. Neither the differences in fOC nor in CEC seem to have a noticeable effect. Barron et al. (2009) found considerable differences for the log KOC

values of atenolol and metoprolol between two sorbents (soil and sludge). Further, they report similar to Hyland et al. (2012), and Schaffer et al. (2012) the limitations of the KOC concept for ionizable compounds. Here, it can be confirmed that KOC is not a suitable parameter for comparing and predicting the sorption behavior of organic cations in different aquifers and sediments, since electrostatic interactions are not considered in this approach (Tolls, 2001; Cunningham et al., 2004; Kah and Brown, 2007; Schaffer et al., 2012). As a consequence, the formally calculated KOC changes with the boundary conditions (e.g., ionic strength, pH). In this study, the log KOC

decreased from 3.2 to 2.7 (experiment A1/A2) and from 3.5 to 3.2 (experiment A3/A4).

3.3.2 Experiments B: Desorption

Desorption with tap water led to similar curves for both sediments (experiments B1 and B3, Fig. 3.3). After horizontally reflecting the desorption curves, their shapes are qualitatively symmetrical compared to the breakthrough curves in experiment A1 and A3, respectively. Therefore, the sorption behavior is likely to be fully reversible.

The quantitative modeling of the desorption curves with CXTFIT under equilibrium conditions led to similar results for R and Kd when comparing sorption and desorption (Table 3.4 and Table B2). The observed higher values for α in case of desorption compared to sorption are expected to be caused by kinetic effects (Rahman et al. 2003, Worch, 2004). As initial condition in the desorption model the columns were assumed to be in sorption equilibrium with atenolol (c(x,t0) = 0.73 for experiment B1 and 0.85 for experiment B3) and as upper boundary condition the columns were fed with water containing no atenolol (c0 = 0).

Afterwards (experiments B2 and B4), the addition of CaCl2 resulted in an increase in atenolol concentration (small peak) for both sediments (Fig. 3.3). Due to the higher competition with Ca²⁺, a new desorption equilibrium is established resulting in an increased desorption. Therefore, strong evidence for cation exchange as the main retardation process for the cationic species of beta-blockers on natural surfaces is apparent.

Fig. 3.3 Experimental and modeled desorption curves of atenolol for sediments S1 and S2 with CaCl2 addition after 53 pore volumes (Ca²⁺ concentrations of 40 and 400 mg L⁻¹).

Concentrations c of desorption curves were normalized with the initial concentration c0

of the previous breakthrough curves. Pore volumes were calculated by normalizing the experiment duration to the ideal breakthrough time of the Cl⁻ tracer.

3.3.3 Experiments C: Different concentrations of atenolol

No significant influence on the transport behavior of atenolol could be observed despite several orders of magnitude difference in atenolol concentrations at identical conditions (Fig. 3.4). R varied between 4.3 for c0 = 500 μg L⁻¹ in experiment C2 and 5.4 for c0 = 1 μg L⁻¹ in experiment C1. Due to the independence from the column bulk properties, the obtained Kd values are virtually equal. Kd was 0.8 L kg⁻¹ for C1 and C2, 0.9 L kg⁻¹ for C3, and 1.0 L kg⁻¹ for C4. As with experiments A1–A4, some biodegradation was observed. In experiment C1 the presence of the atenolol metabolite atenolol acid (Radjenović et al., 2008) was confirmed by LC-MS/MS analysis with atenolol acid as reference substance.

The Kd values are considerably lower compared to experiment A3 and A4 with a similar sediment. This is due to increasing concentrations of inorganic cations (ionic strength) resulting in stronger competition at the sorbent. Therefore, the sorption of atenolol to the available exchange sites decreases.

In contrast, Kd values are equal at constant ionic strength. This implies that even at very high concentrations of inorganic cations the affinity of atenolol to the sorbent is independent from its concentration. At the investigated conditions, the calculated Kd0 of <0.001 L kg⁻¹ indicates that around 99.9% of the observed total sorption is caused by ionic interactions. Thus, the underlying sorption mechanism is not changing and a linear sorption behavior can be expected over a wide concentration range (1–30,000 μg L⁻¹). Hence, the influence of the concentration ratio of atenolol to inorganic cations on the transport of atenolol is negligible at all environmentally relevant concentrations. Absolute concentrations of involved cations are rather important, since these define the competitive system. Therefore, it is postulated that only very high molar atenolol concentrations in the range of the inorganic cation concentrations should have a noticeable influence on the exchange equilibrium and thus on the transport behavior of beta-blockers.

Fig. 3.4 Experimental and modeled breakthrough curves (including tracer tests) for different concentrations of atenolol (C1–C4) and metoprolol (D1, Eq = chemical equilibrium model, nonEq = chemical non-equilibrium model) in sediment S3 at constant ionic strength. Concentrations c of the breakthrough curves were normalized with the initial concentration c0. Pore volumes were calculated by normalizing the experiment duration to the ideal breakthrough time of the Cl⁻ tracer.

Table 3.4 Distribution coefficients and calculated relative proportion of cation exchange.

Exp. = experiment, ρb = bulk density, Kd = observed distribution coefficient (including 95% confidence limits of R and 5% error for ρb and ne respectively ), fOC = fraction of organic carbon, Kd0 = calculated distribution coefficient attributed to non-polar interactions, Kd+ = determined distribution coefficient attributed to cation exchange, Eq = equilibrium model, nonEq = non-equilibrium model

3.3.4 Experiment D: Comparison with metoprolol

At c0 = 500 μg L⁻¹ and comparable concentrations of inorganic cations, the breakthrough curve of metoprolol (Fig. 3.4) showed a significantly higher retardation (R = 10.8) when compared with atenolol (R = 4.3). The obtained Kd of 2.3 L kg⁻¹ for metoprolol in experiment D1 was three times higher than in experiment C2 for atenolol (Kd = 0.8 L kg⁻¹). Biodegradation was not detected (100% of c0 were retrieved).

The higher sorption of metoprolol compared to atenolol is in accordance with the results of Barron et al. (2009) and Ramil et al. (2010) who also found significantly higher Kd values for metoprolol. However, only around 0.15% of the obtained Kd of metoprolol can be attributed to non-polar hydrophobic partitioning (Kd0 = 0.004 L kg⁻¹). Although, partitioning of metoprolol is stronger than for atenolol

(four times higher Kd0 at pH = 8) it can still be neglected for the prediction of transport parameters in aquifers sediments. This is caused by the very low TOC of these sediments while simultaneously the fraction of the cationic species is very high.

Hence, cation exchange is still the controlling sorption process. Despite the similar molecular structure of both molecules, considerable differences remain. The obtained electrostatic contribution of Kd+ = 0.8 of atenolol is significantly lower than for metoprolol with Kd+ = 2.3. This might be caused by several factors, e.g., different molecular geometry and charge density distributions, slightly different pKa values, and uncertainties caused by the applied logKOW-logKOC correlations.