Poner en conocimiento y traslado de la prueba

1.2.1. Adsorption as a post-treatment for catalytic reduction or ozonation

As both catalytic reduction and ozonation did not mineralize the TrOCs investigated in Chapter 2, but only resulted in the formation of a mixture of transformation products, the solutions obtained after these pre-treatments were subjected to activated carbon adsorption. This was done in order to find out what the effect of the different types of pre-treatment is on adsorbability of the transformation products. A comparison with AC adsorption of the parent compounds was performed as well. Mixed results were found, proving that no preceding technology can be considered as being unambiguously superior to the other. Reductive transformation products of diatrizoic acid, atrazine and dinoseb are adsorbed better than their respective oxidative transformation products. On the other hand, bromoxynil’s ozonation products adsorb better than its reduction transformation products. For carbamazepine, both pre-treatments prove to decrease activated carbon adsorption. Note that most differences in AC adsorption removal are relatively small, except for the case of diatrizoic acid, where DABA, the only and final catalytic reduction product of diatrizoic acid, showed a major increase in adsorption behavior.

As a result, the success of applying a sequence of catalytic reduction – AC adsorption depends on the composition of the TrOCs in the water. In most real waters, multiple TrOCs will be present. Therefore, if a combination of either catalytic reduction or ozonation and AC is to be applied, the

choice on the pre-treatment technology will have to be made by balancing out the change in AC adsorption of the different TrOCs present in the feed water.

Note that in this study only 5 TrOCs were tested, because of the limitation of having to use TOC analysis as a measure for adsorption (as oxidation/reduction by-products could not all be determined individually). Due to the limited sensitivity of TOC-analysis, relatively high TrOC concentrations needed to be used, which resulted in the boundary conditions of the solubility limitations of TrOCs. This study therefore only provides a first outline, however, a more thorough investigation involving more solutes is highly recommended. To achieve this, TrOCs with low solubility will inevitably need to be included, and as such, either advances in TOC analysis need to be made, or a full identification and quantification of the reaction mixture products needs to be achieved, so that adsorption can be measured with GC and LC techniques. Neither of these are straightforward. Analysis of low TOC concentrations (below 0.1 µg/L) is difficult due to easy contamination of sample recipients and background carbon in ultrapure water. Alternatively, a full identification and quantification of degradation products would require the synthesis of primary analytical standards of degradation products, or isolation of them from reaction product mixtures. Both are expensive procedures requiring a lot of resources.

1.2.2. The oxidation – adsorption combination in drinking water treatment: reflections Knowing that the difference in adsorption between ozonation products and parent TrOCs is small, the widely used ozonation – GAC adsorption sequence for TrOC removal could be put up for debate. This is because in GAC filters, small solutes such as TrOCs and their associated transformation products are expected to be removed mainly due to adsorption, and not through biodegradation [225, 227, 231]. Using both ozonation and AC adsorption has the apparent advantage that two techniques can partake in TrOC removal, however, if ozonation does not overall benignly alter adsorbability, it may be economically more viable to rely on a well-designed AC adsorption filter only, without a preceding ozonation step. This statement is of course under the assumption that both TrOCs (as parent compounds) and transformation products are undesirable and to be removed.

To date, no legislation about transformation products exists. The current Flemish drinking water legislation dictates that pesticides should be removed to a maximum concentration of 0.1 µg/L per individual compound, which is in fact a norm imposed from an ethical perspective (since pesticides are anthropogenic components, they do not belong in drinking water and thus should not be present in it). This reasoning may be extended to the transformation products as well, as

Note that other anthropogenic molecules, such as pharmaceuticals and industrial products, are not included in the legislation either. From an ethical point-of-view, also these other anthropogenic components and their transformation products are undesirable in drinking water.

In addition, by choosing not to apply a preceding ozonation, no unknown transformation products are generated, and therefore the operational time of GAC filters can be determined by monitoring breakthrough of TrOCs. If TrOCs are first transformed, unknown transformation products may have unknowingly broken through, before the decision is made to regenerate the GAC.

These unknown oxidative transformation products can sometimes be of concern, as oxidative transformation products may be more problematic from a toxicological point of view [294]. Schlüter-Vorberg et al. (2015) [287], Larcher et al. (2012) [288], Ferrando-Climent et al. (2017) [393], Hamdi El Najjar et al. (2014) [394], Dantas et al. (2011) [395], Gómez-Ramos et al. (2011) [396] and Illés et al. (2014) [397] reported that oxidative transformation products of some TrOCs (e.g. tamoxifen, acyclovir, 17α-ethinylestradiol, paracetamol, propranolol, sulfamethoxazole and ketoprofen) have higher toxicity than their parent compounds. Fatta-Kassinos et al. (2011) [398] wrote a review stating more examples of higher toxicity of oxidative transformation products compared to the parent compounds. On the other hand, other examples where toxicity is decreased have been reported as well [398, 399], sometimes only at high oxidant doses [400]. It may thus be concluded that oxidative destruction of TrOCs can result in both a decreasing and increasing toxicity of transformation products, depending on the solutes present. Considering there is no universal decrease in toxicity after TrOC oxidation, and from the ethical stand-still principle that the quality of water should at least remain at its present level (and not decrease), it may be better not to apply an oxidative pre-transformation process.

Note that also the reduction technologies investigated in this work result in the formation of TrOC- based transformation products, and thus the same ethical considerations as mentioned with oxidative technologies also apply here.

1.2.3. Adsorption as a post-treatment for advanced reduction

Similar as with catalytic reduction, advanced reduction will only find a real breakthrough when it either decreases toxicity of TrOCs more efficiently than AOPs (which would be in compliance with the ethical stand-still principle), or if it transforms TrOCs into products which are better removable in a subsequent unit operation, such as activated carbon (which would result in an overall higher removal of TrOCs and associated transformation products). As highlighted in paragraph 1.1.2, the

transformation products of advanced reduction are different than those of catalytic reduction, therefore the AC adsorbability of ARP-products is yet to be determined.

1.2.4. Molecularly imprinted polymer adsorption

In activated carbon adsorption, not only target molecules (such as TrOCs) are being removed, but also harmless dissolved organic matter. Therefore, in Chapter 5, molecularly imprinted polymers, being a novel adsorbent type, were tested for selective adsorption of TrOCs in real water matrices. Target TrOC removal using selective MIP adsorbents is featured by fast adsorption kinetics, in which initially a primary selective adsorption is taking place (i.e. through 1-on-1 matching of specific cavities and their target TrOCs), followed by a slower levelling out adsorption taking place at non-selective binding sites. MIP adsorption equilibrium is being reached much faster than powdered activated carbon. The fast initial adsorption accounted for 72% of metoprolol equilibrium on β-blockers MIP, and 93% of ketoprofen equilibrium on NSAIDs MIP, both being reached within 12 minutes. Ultimately, complete equilibrium is reached after 4 and 8 h for metoprolol and ketoprofen on β-blockers MIP and NSAIDs MIP, respectively. Since a large portion of the total equilibrium is reached in such a short time, i.e., since there is such fast adsorption, MIP adsorbents can be applied in smaller reactors compared to PAC.

The selectivity of MIP adsorbents, however, also has its limitations. In a mixture of 33 TrOCs, target hydrophilic TrOCs (i.e. salicylic acid and atenolol for NSAIDs MIP and β-blockers MIP, respectively) do not adsorb, while non-target hydrophobic TrOCs do show adsorption on MIPs. Also structurally similar solutes to the target solutes are removed by MIP adsorbents, which decreases the adsorption capacity for the target solute. Finally, some anionic non-target solutes also adsorb on MIPs. This shows that some at least some non-selective binding of small solutes occurs, who will thus compete with target solutes for adsorption sites.

Adsorption of both target and non-target TrOCs decreases in different (non-ultrapure water) matrices, however the extent to which this occurs depends on the MIP type and the DOM type. The β-blockers MIP suffers less from competition by background DOM, which is likely due to a difference in nature of the selective adsorption sites compared to those of NSAIDs MIP. MIPs are produced using a target solute(s) (or a molecule(s) structurally similar as the target solute(s)) as a template. NSAIDs all incorporate a carboxylic acid functional group, while the functional structure of β-blockers is a hydroxyl functional group with adjacent secondary amine group, followed by a propyl group. The latter structure is much more complex than the former, and therefore less commonly found in DOM in real waters such as surface water and wastewater effluent. As a result,

DOM can compete more easily with NSAIDs for selective binding sites on NSAIDs MIP than for β-blockers on the β-blockers MIP. This is also supported by comparing MIP adsorption to PAC adsorption, where only the β-blockers MIP is less influenced by competition with DOM.

As such, the selectivity of MIPs depend on the MIP itself, as well as the properties of the different TrOCs present and competing matrix organic matter.

Even though adsorption kinetics are high, and selective removal is possible, one major drawback of MIPs are their high production costs. According to the manufacturer, MIPs cannot be produced below € 900 per kg [401], which is much higher (on average a factor 1000) than the market price of PAC at € 600 – 4500 per ton [402]. As such, it is unlikely that MIPs will be applied for water remediation of TrOCs.

1.3. Comparison of the different experimental technologies applied in this work