Efficient and reversible CO 2 capture in bio-based ionic liquids solutions Journal of CO2 Utilization

Choline/amino acid-based ionic liquids were synthetized via ionic metathesis and their CO 2 absorption perfor- mances evaluated by employing different experimental approaches. In order to overcome any viscosity-related problem, dimethyl sulfoxide (DMSO) was employed as solvent. IL-DMSO solutions with different IL concentra- tions were evaluated as absorbents for CO 2 , also investigating their good cyclability as desirable for real industrial CO 2 capture technologies. 1 H-NMR and in-situ ATR-IR experiments were the toolbox to study the CO 2 chemical fixation mechanism under different experimental conditions, proving the formation of distinct chemical species (carbamic acid and/or ammonium carbamate). In general, these ILs demonstrated molar uptakes higher than classical 0.5 mol CO 2 /mol IL and the capacity to release CO 2 in extremely mild conditions. The possible biological adverse effects were also analyzed, for the first time, in zebrafish ( Danio rerio ) during the development, by assessing for different toxicological endpoints, proving the non-toxicity and high biocompatibility of these bio-inspired ILs.


Introduction
The atmospheric concentration of carbon dioxide (CO 2 ) has increased due to the anthropogenic contribution overstepped 400 ppm in 2015 [1,2]. Fossil fuel combustion for energy production, transportation and industrial processes establishes the main contribution to human emission. Purification of post-combustion gases, storage and conversion/utilization of CO 2 represent a straightforward measure to reduce anthropogenic emissions. Different CO 2 separation techniques are available based on different physico-chemical phenomena: absorption or adsorption, either physical or chemical, membrane or cryogenic-based separation. Besides the working principle, energy consumption, toxicity and operating cost must be taken into account for the technologic implementation [3]. Amine scrubbing process is the most notable and widespread technology: the first process was patented in 1930 [4]. Alkylamine aqueous solutions chemically fix the CO 2 molecule, as shown in Scheme 1. Primary and secondary amines produce ammonium-carbamate species, whereas ammonium-carbonate species are formed from tertiary amine and water (see Scheme 1) [5][6][7]. After the capture, the amine solution is regenerated at 100− 120 • C via water evaporation and the released CO 2 is compressed to 100− 150 bar for sequestration and transportation [8].
The CO 2 release from amine aqueous solutions requires solvent evaporation/condensation, an intensive energy demanding process. Moreover, amine scrubbing has several drawbacks related to the toxicity and the corrosiveness of the sorbent phase. Solving these issues would positively affect both the environmental and the economic impact of the CO 2 capture procedure. Indeed, a lower energy consumption reduces the operational costs and carbon footprint of the entire process, as well as non-corrosive, non-toxic materials are more suitable from the environmental and plant safety point of view.
Ionic liquids (ILs), usually defined as salts with melting temperature lower than 100 • C [9,10], are emerging as promising candidates for CO 2 capture and conversion [11][12][13][14]. ILs can solve some critical issues of amine-based aqueous systems, as they feature negligible vapor pressure, non-corrosiveness and high thermal stability [15][16][17][18][19]. Beyond physical absorption (which becomes relevant at high pressure), the ILs affinity towards CO 2 can be increased by chemically functionalizing the anion and/or the cation, i.e. by introducing an amine group or other basic moieties [20]. Amine-based and amino-tethered ILs capture CO 2 more efficiently through the formation of carbamate species, with a ILs:CO 2 2:1 stoichiometry [20,21], thus also activating the molecule and opening the path for further reactivity [22,23]. Nonetheless, despite the epithet of "green solvent" [24], ILs commonly tested for CO 2 capture are based on imidazolium and pyridinium cations and fluorinated anions, that hamper their biodegradability and biocompatibility [25,26]. In a green perspective, the choice of non-toxic and biocompatible cations/anions represents a key step to develop a sustainable CO 2 capture process based on ILs. In this scenario, the combination of choline (Cho), a non-toxic, essential nutrient cation, and amino acids (AAs, proteins building block, in the anionic form) could drive to the synthesis of an environmental-friendly class of ILs. In particular, AAs are very convenient anions for a CO 2 capture application, owing to the coexistence of an amine moiety (i.e. a specific interaction site for CO 2 ) and of a carboxylic anion (that enhances the affinity with the sorptive) [27][28][29], as practically demonstrated by some literature works. Zou et al. tested the capture performance of a choline prolinate -polyethylene glycol 200 ([Cho][Pro]-PEG200) system at a pressure of 1.1 bar, yielding to a 0.6 CO 2 -IL molar ratio [30]. Lu and co-workers studied aqueous solution of choline glycinate, alaninate and prolinate up to 15 bar of CO 2 , obtaining CO 2 -IL molar ratios up to 2 and conversion of the carbamate to carbonate, due to the presence of water as solvent and the high pressure involved [31]. More recently, our group revealed closely equimolar CO 2 absorption capacity by some AAs -based ILs (synthetized using Choline as cation and glycine or proline as anions) in dimethyl sulfoxide (DMSO) solution, also investigating the role of the IL concentration on the capture performances [32]. Furthermore, materials based on Cho and AAs (in principle achievable from renewable feedstock) [33,34] have demonstrated good biocompatibility and biodegradability [35][36][37][38].
Despite the interesting physico-chemical [39][40][41] and toxicological [35][36][37][38] properties, the employment of these ILs have not spread yet. A first limitation to their employment is the use of choline hydroxide in the usual synthetic procedure. This precursor in expensive, dangerous (it is a corrosive strong base) and difficult to handle. The titration synthesis of neutral AAs by choline hydroxide provides high yields (>90 %) at long reaction times [35][36][37][38], involving the aforementioned chemical threats. These factors limit the scalability of this synthetic strategy. In a recent study, we overcome this issue by applying a synthetic approach based on ionic metathesis, suitable for producing higher amounts of choline-based AAILs with a safer procedure [32,42]. The only drawback is the lower purity of the final product due to the presence of halide salts as impurities (< 5 wt% of KCl). However, regarding the application of these ILs in CO 2 capture, this contamination does not seem to affect their performances [32]. Another general drawback of ILs is their high viscosity, which, unfortunately, further increases upon CO 2 interaction [43]. An excessive viscosity prevents gas diffusion into the liquid, negatively affecting the overall capture performances. As a solution, the dilution with proper solvents decreases the viscosity and, at the same time, can also improve the overall absorption capacity thanks to possible solvent-IL synergy and enhanced transport properties [13,14].
In the present work, six different choline-based AAILs, [Cho][AA] ILs, were synthetized via ionic metathesis and their CO 2 absorption capacity evaluated by a multi-technique approach in order to evaluate if the presence of different functional groups in the AA anion affects the CO 2 capture performance. To overcome any viscosity-related issue, dimethyl sulfoxide (DMSO) was selected as solvent, thanks to its high boiling point (189 • C) and its polar and aprotic character. IL-DMSO solutions with different IL concentrations were evaluated as absorbents for CO 2 (also investigating their cyclability). The effective CO 2 chemical fixation was proved by 1 H-NMR and in-situ ATR-IR experiments. Furthermore, potential in vivo developmental toxicity induced by one of the synthesized ILs has been evaluated on zebrafish (Dario rerio), as emerging vertebrate in vivo models for nanotoxicity screening, to stress the biocompatibility of this new class of ILs [44][45][46][47]. Here, the effects of this ionic liquid were analysed for different toxicological endpoints, including cardiac toxicity, behavioral and possible growth perturbations in zebrafish embryos/larvae. Notably, [Cho][AA] ILs were found to be non-toxic and IL solutions in DMSO demonstrated a molar uptake >0.5 mol CO 2 /mol IL and the capacity to release CO 2 in mild condition, i.e. avoiding the solvent evaporation.

Materials and syntheses
The ILs were prepared by ionic metathesis and purified as previously reported and described in detail in the Supporting Information (Section S1 and Scheme S1) [32,42] The syntheses were confirmed by means of ATR-IR ( Fig. S1 and Table S1) and 1 H-NMR (Figs. 3, S2 and Table S2) spectroscopies. The residual KCl content (usually lower than 3 wt%) was evaluated by TG analysis (Fig. S3 and Table S3). AA salts were also prepared as reference materials, as described in Section S1 of Supporting Information. When needed, [Cho][AA] ILs were dissolved in DMSO (supplied by Merck, purity ≥ 99 %) with different concentration. Solutions were prepared as follows: the IL was weighted in a glass vial, and then the DMSO was added, by adjusting its quantity in order to obtain the desired concentration. About 5 g of solution were produced for each concentration. Viscosities and gravimetric densities of AAILs and their DMSO solutions were assessed and listed in Table S4.

CO 2 absorption
The absorption properties of the DMSO-IL systems were evaluated by means of two different experimental setups. A gravimetric approach was employed to measure the CO 2 capacity, whereas multiple absorption/ desorption cycles were studied in a custom-made batch reactor with an IR CO 2 sensor to obtain information about the regenerability of the systems.

Gravimetric measurements
The CO 2 absorption was quantified using a gravimetric method. Different IL dilution in DMSO were tested: 50, 33, 20 and 12.5 wt%. Approximately 3 mL of solution were poured in a batch reactor (~4.5 mL), purged for 10 min with N 2 (30 mL/min) and weighted. Then, CO 2 (30 mL/min) was bubbled until no mass increase was observed. The captured CO 2 was calculated by mass difference before and after CO 2 contact, considering the reactor headspace contribution. Two quantities were calculated: CO 2 loading, defined as percentage of captured CO 2 over the total mass of DMSO-IL solution, and the molar efficiency, defined as the molar ratio between captured CO 2 and the amount of IL. The solvent evaporation was previously evaluated on pure DMSO and Scheme 1. Chemical reaction of CO 2 with: (a) a primary/secondary amine; or (b) a tertiary amine in the presence of water. estimated to be lower than the balance sensitivity. The error was estimated by propagation starting from instrument sensitivity.
The adsorption capacity of pure ionic liquids has not been evaluated due to their high viscosity, which further increases when they react with CO 2 (until gel and foam formation), hampering the CO 2 diffusion into the bulk liquid phase. The first preliminary tests pointed out a negligible increase in weight also due to the loss of IL stripped by the gas flow.

Adsorption/desorption cycles
Multiple absorption and desorption cycles were carried out in a custom-made reactor provided by HEL Group (schematic is reported in Fig. S4). In a typical experiment, ~5 mL of DMSO-IL solution were used, according to the following steps: (i) purging with a N 2 flow (100 mL/ min) for 10 min; (ii) absorption with a synthetic flue gas (20 %v CO 2 in N 2 [48] -100 mL/min) until saturation (inlet and outlet gas stream had the same composition); (iii) desorption in N 2 flow (100 mL/min) increasing the temperature to 80 • C at 10 • C/min (this temperature was kept until all the CO 2 is released, i.e. when the CO 2 is no more detected in the outlet gas stream); (iv) cooling in N 2 flow (100 mL/min). Absorption/desorption cycles lasted similarly, approximately ~90 min. The steps from (ii) to (iv) were repeated 10 times.

IR spectroscopy
Attenuated total reflection infrared spectroscopy (ATR-IR) was employed to characterize pure ILs and to evaluate the interaction between CO 2 and DMSO-IL solutions. Measurements were carried out on a Bruker Invenio R Fourier transform spectrophotometer equipped with a mercury-cadmium-telluride (MCT) cryogenic detector. The spectra were acquired by accumulating the 32 scans (64 for the background spectrum) in 4000 -600 cm − 1 range with a resolution of 2 cm − 1 .
An in situ IR experiment was specifically designed to mimic real operating condition: a synthetic flue gas mixture was employed (20 %v CO 2 in N 2 ) was adopted, whereas pure N 2 was chosen as inert gas. The interaction between the synthetic flue gas and DMSO-IL solutions was studied by using an Axiom-Hellma TNL 130H multiple reflections ATR cell designed for liquids analysis. The cell is equipped with an AMTIR-1 refractive element and the thermal control was ensured by a recirculating thermostatic bath. The gas composition and flow rates were controlled by means of a modified version of the setup described in ref. [32] (scheme reported in Fig. S5). In a typical experiment, 3 mL of 12.5 wt% IL-DMSO solution were injected in the cell and undergone to (i) purging, (ii) primary CO 2 absorption, (iii) thermal desorption, (iv) cooling and (v) secondary CO 2 absorption. In detail, the solution was: (i) purged with N 2 (100 mL/min) for 10 min to desorb water and other volatile impurities; the temperature was initially set to 25 • C; (ii) exposed to the synthetic flue gas (50 mL/min) at 25 • C; (iii) purged with N 2 (50 mL/min) while increasing the temperature to 80 • C with a 1 • C/min rate; (iv) cooled down to 25 • C under N 2 flow (20 mL/min); and (v) re-exposed to the synthetic flue gas (50 mL/min) for a second absorption step. Spectra were acquired continuously across each step.

Zebrafish maintenance
Adult zebrafish were maintained as previously reported [49]. Briefly, zebrafish were kept at 14h:10 h light-dark cycle and a temperature of 28 • C and were fed three times a day.

Developmental toxicity evaluation of ILs
Embryos at 4 h post-fertilization (hpf) were selected and placed in 24 well-culture plates in the medium [50]. Embryos were incubated at 26 ± 1 • C with four concentrations of [Cho][Ser] aqueous IL test solutions (10, 50, 100, and 200 ppm), and medium as a negative control. Survival and hatching rates were measured daily, while the frequency of movements and the heartbeat rate were calculated in zebrafish larvae at 72 hpf. All the analyses were performed by using a stereomicroscope equipped with a CCD camera. All the experiments were done in triplicates. All animal experiments were performed in full compliance with the revised directive 2010/63/EU.

Statistical analysis
All data were presented as mean ± S.D. Differences among the treatments were analysed by one-way analysis of variance (ANOVA) in combination with Holm-Sidak post hoc. A difference between the treated and the control group was considered to be statistically significant at p<0.01.

CO 2 absorption capacity
The CO 2 absorption capacity of all [Cho][AA] ILs was assessed in DMSO solution at different concentration, by means of a gravimetric setup (described in detail in the experimental section). Tests were performed by bubbling pure CO 2 (1 atm) directly inside the solutions until saturation. The results (reported in Fig. 1 Table S5) were similar for all the amino acids tested as IL anions, however all the [Cho]

and in detail in
[AA]-DMSO solutions exhibited absorption performances significantly higher compared to literature data of pure DMSO (which report values of about 0.5 wt% at 1 atm) [51]. It is worth noting that by decreasing the IL concentration in DMSO, the solution absorption capacity decreases, whereas the molar efficiency increases. This behavior is in line with previous results reported for [Cho][Gly]-DMSO and [Cho][Pro]-DMSO solutions [32]. Indeed, the higher distance of the ionic couples in less concentrated solutions affects the absorption mechanism, favoring the formation of carbamic acid and, at the same time, hindering the proton transfer involved in the ammonium-carbamate generation. In general, the various ILs exhibit not so different molar efficiencies ( Fig. 1, part 4 ], compared to the chemisorption of functionalized ILs. Among the functionalized ILs, the absorption capacity (AC) spreads approximately from 4 to 15 wt%, while the molar efficiency (ME) from ~ 0.2 to 0.9 mol/mol. Even though the set of ILs recently studied by Xiong et al. shows notably high absorption capacities, our samples have significantly higher MEs and, at the same time, the great advantage of good biocompatibility.
The 12.5 wt% concentration, exhibiting a slightly higher molar efficiency in previous gravimetric analyses (Fig. 1), was selected to test the cyclability of all [Cho][AA] ILs aiming to assess the feasibility of cyclic absorption/desorption in a demonstrator unit, simulating the experimental conditions of an industrial CO 2 capture technology. In this experimental setup, the adsorption capacity reflects the absorption and desorption rates of the IL solutions, rather than the real thermodynamic equilibrium, as in the gravimetric measurements. As reported in Fig. 2    absorption capacity (AC, g(CO 2 )/g(IL) wt% and g(CO 2 )/g(solution) wt% in brackets), and molar efficiency (ME, mol(CO 2 )/mol(IL)) for different ILs. Presence of solvent, concentration (wt%), temperature ( • C) and pressure (bar) employed are reported in brackets, in the first column. Within each block, ILs are reported in order of increasing ME. [Pro] solution shows the highest relative performance during the last cycles (around 85 %), despite its lower CO 2 uptake, being in some way the best compromise between the capacity to capture CO 2 and the stability along the absorption cycles. It is worth highlighting that the IL containing the phenyl moiety exhibits the best absorption performances during the first cycle, but its absorption capacity decreases more during the cycles. In contrast, the IL containing a heterocyclic moiety with a secondary amine ([Cho][Pro]) possesses a lower absorption capacity, but it clearly maintains its capture performance during the cyclic experiments. Instead, the presence of a secondary amine inserted in a linear aliphatic chain does not seem affecting the capture performance and stability of [Cho] [Sar].
In general, we can conclude that a direct comparison of the cyclic experiments with the gravimetric results is not straightforward, being the CO 2 absorption/desorption equilibria affected by different parameters, among which the IL concentration, the viscosity changes, the surface tension and the CO 2 diffusivity in the solution [57].

Assessment of the absorption mechanism
The CO 2 absorption process into a bulk [Cho][AA] IL is associated with a real chemical reactivity occurring between the CO 2 molecule and the amine group of the amino acid anion. Therefore, beside the evaluation of the absorption capacities of the DMSO-IL systems, we also investigated their interaction mechanism with CO 2 at a molecular level by means of 1 H-NMR and in situ ATR-IR spectroscopies.

NMR spectroscopy
In 1 H-NMR spectra collected upon interaction of the ILs/DMSO-d 6 solutions with CO 2 (in similar conditions to those adopted in the gravimetric absorption measurements, i.e. employing a pure CO 2 stream), the formation of carbamate/carbamic acid species was depicted by a remarkable de-shielding of the signals of the α and β protons to the When the [Cho][Ser] solution (12.5 wt%) interacts with pure CO 2 , the α proton undergo a downshift from 2.91 ppm to 3.42 ppm, overlapping to the choline signal, while the well-defined β protons in the pending group at 3.33 and 3.27 ppm shifts to 3.56 ppm (Fig. 3a). [Cho][Sar] IL solution exhibits a similar behavior upon CO 2 absorption, both the α protons pattern and the sarcosinate N-Methyl group signal undergo a downshift from 2.77 to 3.56 ppm and from 2.22 to 2.67 ppm respectively (Fig. 3b)

In situ IR spectroscopy
The CO 2 capture and release were monitored by means of in situ ATR-IR spectroscopy, under experimental conditions close to the cyclic absorption measurements, in order to identify the chemical species generated by the CO 2 reaction with [Cho][AA] ILs. Room temperature absorption and following desorption at 80 • C were monitored by in situ ATR-IR experiments using 12.5 wt% IL-DMSO solutions of all the synthetized [Cho][AA] ILs. Spectra collected before and after CO 2 interaction and after CO 2 desorption at 80 • C are compared in Fig. 4  (containing a primary or a secondary amine moiety in the AA, respectively) are reported in the main text. The spectra of the remaining ILs are available in Fig. S6 and the detailed summary of the spectral modification generated by the absorption of CO 2 in the different IL-DMSO solutions is reported in Table S6. All reported spectra were collected in different stages of the adsorption/desorption process (as reported in the legend of the figures) after reaching the equilibrium (i.e., when spectral modifications were no more observable).
Spectra of [Cho][Ser]-DMSO 12.5 wt% solution are reported in Fig. 4a. Upon CO 2 absorption (i.e. passing from the black to the red curve), different spectral modifications occur. They are related to the reaction paths reported in the lower part of Fig. 4, i.e. to the formation of ammonium carbamate (i) and carbamic acid (ii) species, whose IR bands are labelled in Fig. 4 by circles and squares, respectively. In particular, the formation of the ammonium moiety is testified by the appearance in the red spectrum of [Cho][Ser]-DMSO of the shoulder at 1650 cm − 1 and of the band at 1495 cm − 1 , ascribed to the asymmetric and the symmetric bending modes of the NH 3 + ion. A further confirmation arises from the changes in the vibrational modes of carboxylate species: a +15 cm − 1 shift of the OCO-asymmetric stretching is indeed evident, suggesting the formation of new carboxylate species (i.e., present in the carbamate moiety). The formation of carbamic acid, instead, is testified by the growth of an intense band at 1700 cm − 1 , ascribed to the stretching mode of its carbonyl group.  Fig. S6).
In contrast, the [Cho][Sar]-DMSO 12.5 wt% solution (see Fig. 4b) spectroscopically behaves in a slightly different way. Indeed, in the presence of [Cho][Sar], the protonation occurs on a secondary amine moiety, giving rise to NH 2 + species when the ammonium carbamate forms. The asymmetric and symmetric NH 2 + bending modes appear at 1630 and 1490 cm − 1 , respectively. The formation of ammonium carbamate species is further confirmed by the evolution of the OCOstretching vibrations: upon CO 2 interaction the asymmetric mode undergoes an upward shift from 1595 to 1610 cm − 1 while it decreases in intensity, partly overlapping to the signal of protonated amines. The formation of carbamic acid is instead confirmed by the appearance of a peak at around 1700 cm − 1 , ascribed to the stretching mode of the carbonyl moiety. A similar behavior is observed for [Cho][Pro] (containing an AA with a heterocyclic secondary amine) [32]. [AA] IL. Indeed, all the aforementioned signals disappear, testifying that, during the ATR-IR experiments carried out in situ, the CO 2 captured in the form of carbamic acid and ammonium carbamate species is totally released at 80 • C, without any solvent evaporation. In situ ATR-IR measurements proved the capacity of CO 2 to effectively react with all the considered amine moieties.
In particular, two possible reactions paths were detected: the 1:2 reaction of two amine functional groups (deriving from two distinct IL molecule) with one CO 2 molecule, forming an ammonium carbamate couple (reaction path (i) in Fig. 4), and the 1:1 reaction of a single amine group with one CO 2 molecule, producing carbamic acid (reaction path (ii) in Fig. 4). The gravimetric absorption measurements proved that the overall [Cho][AA] IL:CO 2 stoichiometry (determined by the molar efficiency) varies according to the [Cho][AA] IL concentration in the DMSO solution. Now, after the spectroscopic identification of two distinct reaction paths, we can infer the IL concentration likely determines the preferred absorption mechanism, occurring via the "ammonium carbamate route" in the more concentrated solutions or preferentially through "the carbamic acid route" in the less concentrated ones.
We further checked the cyclability of the absorption process on the [Cho][Ser]-DMSO 12.5 wt% solution by means of in situ ATR-IR spectroscopy, by performing a secondary absorption run (always at room temperature), after the primary desorption step at 80 • C. As reported in Fig. S7, spectra collected in situ immediately after the first and the second absorption cycle, show negligible differences. This result again highlights the excellent cyclability of these IL-DMSO solutions.
These last results seem to be in contrast with the data of the multiple absorption/desorption cycles collected using the custom-made reactor (to mimic the experimental conditions of a real CO 2 capture technology) and reported in Fig. 2. For this reason, we decided to collect ATR-IR spectra of one of the IL-DMSO solution immediately after the synthesis (fresh) and after ten CO 2 absorption/desorption cycles (used: 10 cycles) in the custom-made reactor, to explain the decrease of the absorption performances observed after several cycles employing this setup. The [Cho][Phe]-DMSO solution was selected due to its more pronounced decrease of the CO 2 absorption capacity during the cyclic tests. Spectra are reported in the Supporting Information (Section S3.3, Fig. S8). Before the ATR-IR measurements, the 10th cycle has been followed by a final desorption step at 80 • C, until CO 2 was no more detected in the [Sar]-DMSO solution 12.5 wt% before CO 2 absorption (black curves), after CO 2 absorption (red curves) and after desorption at 80 • C (blue curves). Dark red circles (•) and squares (◼) highlight relevant spectral modifications related to the formation of ammonium carbamate and carbamic acid species, respectively. Lower part: chemical reactions between amines and CO 2 forming (i) ammonium carbamate and (ii) carbamic acid. The spectrum of bare DMSO is reported as reference (grey dotted curve). outlet gas stream. It means that, if the absorption process is totally reversible, the spectra collected on fresh sample (black curve) and on the solution after the 10th absorption/desorption cycle (red curve) should coincide. Fig. S8 reports the spectra of the [Cho][Phe]-DMSO solution at the concentration of 12.5 wt% in the 1800-1100 cm − 1 spectral range, where the bands generated by the reaction with CO 2 are mainly located. The spectrum collected after 10 cycles (red) does not show any spectral evidence of possible degradation phenomena of the IL and exhibits only the bands generated by the chemical absorption of CO 2 . The presence of the signals due to the species generated by the chemical reaction of CO 2 with the IL proves that the specific conditions used in cyclic experiment performed in the custom-made reactor do not allow a complete regeneration of the IL-DMSO solution upon several absorption/desorption cycles. Despite these results, the in situ ATR-IR measurements clearly highlighted the great potentiality of these bio-based ionic liquids solutions that can ideally release all the absorbed CO 2 in mild conditions.  Fig. 5. In particular, the hatching and survival rates were monitored every 24 h. Following the exposure with IL from 10 ppm up to 200 ppm, the survival rates showed a profile time and concentration-dependent with no relevant decrease during the temporal window analysed (Fig. 5a). At the highest concentration investigated (200 ppm) and after 120 hpf, the value of survival was up to 95 %. In addition, the ability to successfully hatch (hatching rates) of the treated groups showed negligible reduction compared to the control group (Fig. 5b). The treated embryos hatched in the normal temporal window (between 48 and 72 hpf). In fact, at 72 hpf, 95 % of embryos treated with the highest concentration of [Cho] [Ser] hatched. In accordance with the OECD guidelines [58], the profiles and trends of the hatching and survival rates of zebrafish treated with [Cho] [Ser] indicated that the investigated IL did not have adverse effects on the embryogenesis of zebrafish.

In vivo toxicity assessment
To further investigate the effects of the IL on zebrafish, we measured the heartbeat rates and frequency of movements of treated larvae at 72 hpf. The heartbeat rates of larvae exposed to the [Cho][Ser] showed no significant decrease or increase compared to the control groups (Fig. 5c). Similarly, the frequency of movements of 72 hpf larvae exposed to [Cho] [Ser] (Fig. 5d), presented no perturbations in comparison with the control samples. The values of heartbeat rate and frequency of movements indicated that [Cho][Ser] did not influence the cardiac and swimming activities of treated larvae, further confirming that the IL did not affect the embryogenesis.

Conclusions
In this work, we presented the complete advanced characterization of different amino acid-based ILs (   solution, containing a heterocyclic moiety with a secondary amine, exhibited the highest stability along the absorption cycles, loosing just the 15 % of its absorption performance during the first ten cycles. Further studies are needed to really understand the main reason of this peculiar behavior and to explain how the presence of different functionalities in the IL affects the final absorption performances. The CO 2 capture and release mechanisms were studied in detail by means of 1 H-NMR and in situ ATR-IR spectroscopies under different experimental conditions. 1 H-NMR spectroscopy confirmed the neat prevalence of a unique species identified with carbamic acid when the sorbents are exposed to a pure CO 2 stream until saturation. ATR-IR spectroscopy, performed employing a synthetic flue gas stream (i.e. in the presence of diluted CO 2 ) proved the formation of both carbamic acid and ammonium carbamate species, justifying the CO 2 : AAIL stoichiometry in the 0.5-1 range observed in the gravimetric experiments. It is evident that the CO 2 relative pressure in the gas stream and the different experimental conditions could be responsible for the preferable absorption pathway giving rise to distinct species (carbamic acid and/or ammonium carbamate). Indeed, the high CO 2 pressure employed in NMR experiments probably promotes the formation of a single product (carbamic acid).
Another important result concerns the CO 2 temperature release of the [Cho][AA] IL-DMSO solutions. Indeed, it is worth noting that, compared to the classical aqueous amine solutions (requiring temperatures higher than 100 • C to achieve a complete CO 2 release), these ILs totally desorbs CO 2 at milder conditions (the total release of CO 2 occurs at 70− 80 • C, as clearly proved by in situ ATR-IR experiments).
Finally, the present study reported, for the first time, the in vivo evaluation of the toxicological profile of a [Cho][AA] IL in vertebrate systems, demonstrating the non-toxicity and high biocompatibility of [Cho][Ser] in zebrafish during the development. More in general, the results reported in this work proved how these bio-inspired ILs are very promising alternatives to the classical amine aqueous solutions, mainly thanks to their significantly low CO 2 release temperature, good regenerability and high biocompatibility.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.