3.1 Material design and characterization
As mentioned, with the purpose to obtain bio-derived adsorbents, the amine mediated ring opening reaction of di- and tri-glycidyl ethers has been exploited to cross-link a commercial maltodextrin, giving cationic polymer structures as product. Even though the epoxy ring-opening reaction is reported to occur already at 50°C in alkaline water solutions, the addition of tertiary amine during the synthetic step allows to obtain polymers characterized by the presence of positively charged pendants, since the nucleophilicity displayed by amines allows the latter to be kinetically favoured compared to hydroxyl species (Rodriguez-Tenreiro et al. 2006). This approach allows also to decrease the polymerization reaction time, since the amine introduced let a faster production of alkoxide intermediates, responsible of the subsequent generation of polymer chains. An initial interaction, occurring between DABCO and the linker (Figure 1A), was reported to generate alkoxides species, which subsequently led to the actual growth of polymer chains (Figure 1B-E) (Cecone et al. 2021). Once formed, the alkoxides could react with either other epoxide rings belonging to the linker or hydroxyl functions from GLU2. In the first case, the reaction was associated mostly with a propagation of the alkoxide species, but a negligible molecular weight increase if related to unreacted linker molecules. On the other hand, the alkoxides generated on the backbone of GLU2 molecules which could further react with other epoxides rings, were considered to be of great impact to the increase of the molecular weight, being GLU2 characterized my molecular weight of approximately 300k Da (Stijnman et al. 2011). As a result of the reactive path described, a cationic polymer network was reported as product. As a confirmation of that, it was possible to recover a polymer product from all the synthetic conditions chosen for this work. Said so, the first characterization was carried out by comparing the amount of product recovered from each synthesis. The amount of polymer obtained after the purification and drying process was expressed as mass balance, considering the weight of the final product with respect to the theoretical weight, equal to the sum of GLU2, DABCO and the cross-linker. The highest mass balance was observed for GLU_TTE which was equal to 88.1 %, followed by GLU_NGDE and finally by GLU_BDE, 86.3 % and 85.8 % respectively. As reported in Figure 2, the polymer products were characterized by the presence of polydisperse granules whose size was approximately between a few tens to twohundred microns. The morphological characterisation also revealed the presence of smooth external surfaces. Further, to evaluate the thermal stability, a thermogravimetric analysis was carried out for each sample. As reported in Figure 3A, the thermogravimetric profiles of all samples were characterized by a first weight loss occurring approximately up to 150°C which was related to the volatilization of the water adsorbed on the samples. Subsequently, between 250°C and 450°C GLU_BDE profile displayed a single step decomposition process, whereas GLU_NGDE and GLU_TTE profiles indicated a multistep decomposition process. Subsequently, a carbon residue stable up to 700°C was obtained, corresponding roughly the 20% of the initial weight. Also, the highest Tonset was displayed by GLU_BDE which was equal to 302°C, while GLU_NGDE and GLU_TTE displayed a Tonset equal to 287°C and 282°C, respectively. Afterwards, the presence of quaternary ammonium functions within the polymers structure was studied via FTIR-ATR. As reported in Figure 3A, all the samples display a peak cantered approximately at 1590 cm-1, suggesting how the aforementioned functions were displayed over the network. The presence of nitrogen atom composing the polymer structure was further confirmed by the elemental analysis of the samples. The highest amount of nitrogen was detected in GLU_TTE and resulted equal to 1.1%, followed by GLU_BDE corresponding to 0.8% and by GLU_NGDE equal to 0.6%. Furthermore, what obtained was supported by ζ-potential analysis where GLU_TTE displayed a ζ-potential equal to 13.8 ± 1.7%, GLU_NGDE a value of 11.4 ± 1.3% and GLU_BDE a value of 10.4 ± 1.2%. What observed so far, allowed to demonstrate how the presence of tri-functional linkers such as TTE, compared to di-functional ones like BDE and NGDE, resulted in higher amounts of product characterized by higher cationic features. This aspect has been related to the capability of tri-functional linkers to act both as a linker between two GLU2 molecules, and as cationic pendant via reaction with DABCO, by exploiting the remaining epoxy function. On the contrary, di-functional linkers can be present only as a linker or as a pendant. What hypothesized was also supported by the swelling capability which resulted approximately 1400 % in the case of GLU_BDE, 1300 % for GLU_NGDE, and 1100 % in the case of GLU_TTE. Being TTE capable to generate a higher number of cross-links, its structure resulted less prone to adsorb solvents, in this case water.
3.2 Evaluation of adsorption performances
The obtained cationic polymers were subsequently studied as novel adsorbents for the removal of anionic pollutants from water solutions. The study has been performed by choosing nitrates, sulphates, and phosphates as case-study and by testing the adsorption features against each anion at different concentration and pH values. The adsorption tests have been performed starting from 100, 200, 300, 400, 500 and 1000 mg/L solutions in the case of nitrates and sulphates, while in the case of phosphate the adsorption has been carried out by varying the pH value (11.88, 8.49, 7.04, and 5.25), keeping constant the concentration of phosphorous at 32 mg/L. In the case of nitrates (Figure 4A and 4B), it is possible to observe how the adsorption efficiency changed with the variation of the initial concentration of the pollutant. GLU_TTE displayed higher adsorption performances if compared to GLU_BDE and GLU_NGDE, which gave similar behaviours. By looking at Figure 4A, an overall decrease in the percentage of removal efficiency as the nitrate concentration increases was observed. The reason behind this trend was hypothesized to be related to the number of cationic sites displayed by the adsorbent, which can electrostatically interact with the pollutant resulting in a retention of the latter. Once these sites are saturated by the presence of the corresponding number of anions, the adsorption capacity of the adsorbent will be lower, causing an overall decreasing of the adsorption performances at increasing pollutant concentration. In fact, all the adsorbents displayed the higher percent adsorption capability from 100 mg/L solutions, where the highest value was observed for GLU_TTE (79.7 ± 4.0%), followed by GLU_BDE and GLU_NGDE (62.9 ± 3.1% and 59.4 ± 3.0%, respectively). Furthermore, being the electrostatic interaction between the adsorbent and the pollutant occurring as equilibrium between the free anion and adsorbed anion, considering also that the reported adsorbents display swelling features in water media, the total amount of adsorbed pollutant was hypothesized to be higher starting from higher concentration. As a confirmation, the milligrams of nitrate retained per gram of material increased starting from higher concentration of pollutant, as reported in Figure 4B. In this regard, the higher adsorption performances in terms of milligrams of pollutant adsorbed per gram of adsorbent were observed from 1000 mg/L solutions. Again, GLU_TTE was the most performing system with 38.0 ± 0.6 mg/g, followed by GLU_BDE which resulted in 18.0 ± 0.3 mg/g and GLU_NGDE with 15.0 ± 0.3 mg/g. A similar trend was displayed also in the case of sulphates adsorption tests (Figure 4C). However, by looking at the lowest starting concentrations, higher adsorption capacities have been observed, if compared to nitrate tests. As reported before, also in this case GLU_TTE displayed the best performances with 94.6 ± 4.7% from 100 mg/L solution, while GLU_BDE displayed a value of 64.2 ± 3.2% and GLU_NGDE a value corresponding to 59.4 ± 3.0%. This feature demonstrated how the adsorption efficiency and the adsorption mechanisms in general are affected by the nature of the anion. What observed was related firstly to the higher charge of sulphates if compared to nitrates, allowing the first to be more prone to electrostatically interact with the adsorbent. Also, sulphates display higher molecular weight if compared to nitrates, this second aspect could be the cause of a higher retention of the anion within the polymer network. What described was hypothesized as main reason behind the differences observed. Furthermore, by considering the electrostatic interactions the main phenomena behind the adsorption process, the total amount of sulphate adsorbed was hypothesized to be lower than that of nitrate. This because, given the number of cationic pendants displayed by the adsorbents, sulphates would saturate those sites before than nitrates, displaying four times the charge of the latter for each anion. As a confirmation of that, by looking at the adsorbed amount of pollutants from 1000 mg/L solution, the decrease of performances for GLU_TTE was equal to 23%, 14% for GLU_BDE, and 12% in the case of GLU_NGDE. Eventually, the reported adsorbents were also tested on four phosphate solutions at pH 11.88, pH 8.49, pH 7.04 and pH 5.25, in awareness that phosphates can be found in three different forms depending on the pH value. Between pH 4 and pH 6, phosphates are mostly found as di-hydrogen phosphate species, between pH 8 and pH 11 the predominant form is the mono-hydrogen phosphate, while above pH 12.5 the orthophosphate anion becomes the most represented form (Cantrell et al. 2008; Peng et al. 2018). Having said this, unlike the trends described above, the presence of a maximum centered at pH 8.49 was observed, by comparing the different removal efficiencies. At this pH value, the adsorption resulted quantitative (99.0 ± 0.2%) in the case of GLU_TTE, GLU_BDE resulted in 83.5 ± 1.3%, while GLU_NGDE was equal 87.2 ± 2.1%. This peculiar trend was hypothesized to be mostly affected by two different aspects. Firstly, since the different phosphates anions are characterized by different number of charges, they might display different electrostatic behavior towards the adsorbent. With the above in mind, the orthophosphate was hypothesized to be more prone to interact with the adsorbent if compared to the mono-hydrogen phosphate and to the di-hydrogen phosphate, since the higher the number of charges displayed by the anion, the stronger the interaction generated with the adsorbent. Second, at increasing alkaline condition the presence of hydroxyl anions together with the phosphates species would generate a competition for the interaction with the cationic sites, resulting in lower adsorption performances. As a proof of that, the lowest adsorption values were observed in the most alkaline conditions (pH 11.88). In this case, despite the presence of orthophosphate species, the concentration of hydroxyl anions resulted high enough to hinder the interaction of the pollutant with the adsorbent. On the other hand, in acidic (pH 5.25) and neutral (pH 7.04) conditions, the adsorption performances were affected by the presence of less charged species, if compared to the orthophosphate. However, in this pH range the presence of proton ions in solution did not result in a significant modification of the adsorption performances. The highest adsorption features displayed at pH 8.49 resulted the optimal trade-off between the concentration of hydroxyl ions such that did not hinder the adsorption of the pollutant and the presence of the less protonated ion forms, thus more negatively charged.