Synthesis and Phase Behavior of a Platform of CO2-Soluble Functional Gradient Copolymers Bearing Metal-Complexing Units

The synthesis and characterization of a platform of novel functional fluorinated gradient copolymers soluble in liquid and supercritical CO2 is reported. These functional copolymers are bearing different types of complexing units (pyridine, triphenylphosphine, acetylacetate, thioacetate, and thiol) which are well-known ligands for various metals. They have been prepared by reversible addition–fragmentation chain-transfer (RAFT) polymerization in order to obtain well-defined gradient copolymers. The copolymers have been characterized by proton nuclear magnetic resonance (1H-NMR) spectroscopy, matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry, thermal gravimetric analysis (TGA), dynamical scanning calorimetry (DSC) and cloud point measurements in dense CO2. All the investigated metal-complexing copolymers are soluble in dense CO2 under mild conditions (pressure lower than 30 MPa up to 65 °C), confirming their potential applications in processes such as metal-catalyzed reactions in dense CO2, metal impregnation, (e.g., preparation of supported catalysts) or metal extraction from various substrates (solid or liquid effluents). Particularly, it opens the door to greener and less energy-demanding processes for the recovery of metals from spent catalysts compared to more conventional pyro- and hydro-metallurgical methods.

A greener alternative to the aforementioned methods is the use of a different solvent system, which is dense CO 2 (liquid CO 2 , lCO 2 , or supercritical CO 2 , scCO 2 ). Above its critical point (T c = 31 • C, P c = 7.38 MPa), scCO 2 has properties in between those of liquid and gas, such as high density and high diffusivity. Furthermore, dense CO 2 is a non-polar solvent, unable to solubilize metals; thus, additives are required for metal extractions. So far, different reports have shown the use of low molecular weight compounds, (e.g., dithiocarbamates, beta-diketones, dithizone, perfluorocarboxylic acids) which can act as complexing additives [16,[44][45][46][47][48][49][50][51]. Nevertheless, our group has reported in the last years the possibility to use scCO 2 -soluble metal-complexing polymers to extract metals from solid matrices, without dissolving or destroying the supports [52][53][54][55]. These polymers contain two different types of monomer units, namely CO 2 -philic monomer units ensuring a high solubility of the polymers in scCO 2 , and metal-complexing monomer units enabling binding to the metals. Even if different classes of siloxane-based [56,57] and vinyl alkanoate-based [58,59] polymers have been studied, it is known that compounds having carbon-fluorine bonds are usually soluble at much lower pressures than other compounds [60,61]. The CO 2 -philic monomer unit is represented by 1,1,2,2-tetrahydroperfluorodecylacrylate (FDA), corresponding to a fluorinated poly(1,1,2,2-tetrahydroperfluorodecylacrylate) (PFDA) homopolymer which has an extremely high affinity for dense CO 2 [62][63][64]. The metal-complexing monomer units are usually characterized by polar groups (CO 2 -phobic) able to interact with various metals. The increasing amount of the metal-complexing monomer units has often a high impact on the phase behavior of these polymers in dense CO 2 , decreasing their solubility in this solvent. Another factor that influences the solubility of the copolymers in dense CO 2 is the distribution of the CO 2 -phobic monomer units in the polymer chain: it is reported that gradient copolymers are soluble at a lower CO 2 pressure than block copolymers [65]. A very straightforward method to obtain gradient copolymers is reversible addition-fragmentation chain-transfer (RAFT) copolymerization [66]. In a batch process, the gradient architecture of the copolymer is defined by the reactivity ratio of the monomers (composition drift along the copolymer chain). This reversible-deactivation radical polymerization technique allows the production of gradient copolymer in which one side of the polymer is enriched in the CO 2 -phobic metal-complexing monomer units whereas the other side of the polymer chain is enriched in CO 2 -philic units, while preserving a high solubility of the polymer in dense CO 2 even at high loading of metal-complexing CO 2 -phobic units. Furthermore, the use of RAFT controlled radical polymerization (composition drift within the polymer chains) instead of conventional free radical polymerization (composition drift between the polymer chains) enables to have gradient polymer chains, with similar length and composition, which facilitates the copolymer solubilization in scCO 2 . In this report, we present the preparation and the phase behavior in dense CO 2 of a platform of novel metal-complexing gradient polymers synthesized by RAFT polymerization. These polymers contain different complexing moieties (pyridine, triphenylphosphine, acetylacetate, thioacetate and thiol) which are well-known metal ligands [67][68][69][70][71][72]. The wide range of ligands introduced in the fluorinated polymer chains is usually able to complex a wide range of metals such as Li + , K + , Ag + , Au + , Pd 2+ , Gd 3+ , Hg 2+ and Co 2+ , to name a few [69]. Taking advantage of the RAFT technique, the polymers are further modified to improve their potential complexing abilities through aminolysis reaction, transforming the dithiobenzoate moiety coming from the RAFT controlling agent into a terminal thiol group at the polymer chain end. The synthesis of these gradient polymers and the determination of their phase behavior in dense CO 2 aims at building up a library of CO 2 -soluble metal-complexing polymers ( Figure 1) as candidates to recover different types of metals from industrial and urban wastes.  Synthesis of StySAc is illustrated in Figure S3 in the SI. Vinylbenzyl chloride (20 g, 0.131 mol) and 50 mL of acetone were added to a round bottom flask. The mixture was stirred with a magnetic stir bar for 20 min in an ice bath under N 2 atmosphere. Afterward, potassium thioacetate (1.2 eq., 18 g, 0.157 mol) was added over 30 min and the reaction stirred for 30 min in an ice bath and then 48 h at room temperature. Then, the mixture was filtered on a filter paper and the acetone evaporated under reduced pressure.

Materials and Methods
The obtained yellow oil was solubilized in CH 2 Cl 2 (40 mL) and washed with deionized water (3 × 50 mL). Finally, the organic phase was dried with Na 2 SO 4 and the solvent evaporated under reduced pressure. An amount of 21.4 g of final compound was collected as a brown oil (85% yield). The structure of the product, S-(Vinylbenzyl) ethanethioate (M = 192.28 g/mol), was confirmed by 1 H-NMR analysis, as shown in Figure S4. 2.2.2. Synthesis of Poly(1,1,2,2-tetrahydroperfluorodecylacrylate) (P(FDA)) Homopolymer FDA (40 g, 0.0771 mol), CTA (2.141 g, 0.00842 mol), AIBN (0.415 g, 0.00253 mol) and TFT (42 mL) were added in a Schlenk flask. The mixture was stirred magnetically and bubbled for 40 min with N 2 . Then, the polymerization was started by heating the Schlenk flask in an oil bath at 65 • C. At the end of the reaction, the mixture was let return to room temperature and it was precipitated in 600 mL of pentane 3 times from TFT solution and the polymer was dried under vacuum overnight at room temperature. After drying, 32.8 g of polymer were recovered as a fine pink powder (78% yield).

Aminolysis of the Polymers
As a general procedure for aminolysis of the polymers, polymer (2 g), PPh 3 (5 molar eq. versus aminolyzable group), and TFT (5 mL) were added in a Schlenk flask. The mixture was stirred magnetically and bubbled for 40 min with N 2 . Afterward, piperidine (3 molar eq. versus polymer) was added. The mixture was stirred for 3 h at room temperature. Then, the reaction mixture was precipitated in 60 mL of pentane 3 times from TFT solution and the polymer was dried under vacuum overnight at room temperature.

Cloud Point Measurements
Cloud point measurements were carried out in a high pressure, variable volume view cell equipped with a sapphire window on the end for visual observations. The cell was equipped with a pressure transducer and an internal thermocouple. It was thermostated by a water/isopropanol alcohol mixture delivered by a Lauda RE206 circulating pump. CO 2 is delivered by an ISCO 260D automatic syringe pump. An amount of 55 mg of polymer was weighed and transferred to the cell along with a clean stirring bar at a starting volume of 6.40 mL. Subsequently, the cell was fed with CO 2 at about 25 • C and 10.9 MPa. Then, the cell was heated to 65 • C (taking care to adjust the volume of the cell in order to stay below a pressure of 35 MPa; safety rupture disk at 50 MPa). Cloud points (one-phase/two-phase transition) were obtained by decreasing the pressure of the cell by increasing the cell volume through a hand-driven piston after 15 min of stirring at a given temperature. The cell was cooled by steps of 5 • C down to 25 • C. The uncertainty of the cloud point pressure was ±0.5 MPa.

Nuclear Magnetic Resonance Spectroscopy (NMR)
Chemical structures were determined by 1 H-NMR spectroscopy on a Bruker Avance 400 MHz spectrometer at room temperature. The spectra were recorded by dissolving 10 mg of sample in 0.5 mL of CDCl 3 (StySAc and 4VP-based copolymers), acetone-d 6 (AAEM copolymers), CFC-113 with C 6 D 6 capillaries (other fluorinated (co)polymers).

Thermogravimetric Analysis (TGA)
TGA measurements were performed by placing 10-15 mg samples in an aluminum pan on a Q50-TA Instruments, heating the sample up to 580 • C at 10 • C min −1 under N 2 inert atmosphere.

Differential Scanning Calorimetry (DSC)
DSC measurements were performed by placing 10-15 mg samples in an aluminum crucible on a Netzsch DSC 200 F3 instrument using three heating-cooling cycles from −40 to 150 • C, with 20 • C/min heating rate under N 2 inert atmosphere. Calibration of the instrument was performed with noble metals and checked with an indium sample.

MALDI-TOF-MS Mass Spectrometry
MALDI-TOF mass spectra were performed at the Laboratoire de Mesures Physiques of Montpellier University using a Bruker Rapiflex time-of-flight mass spectrometer and a nitrogen laser for MALDI (λ = 337 nm). The measurements in reflectron positive ion mode were recorded with a voltage and reflector lens potential of 20 kV and 21.5 KV, respectively. Mixtures of peptides were used for external calibration.
The matrix and cationizing agent were trans-2-[3-(4-tert-butylphenyl)-2-methyl-2propenylidene]malononitrile (DCTB) diluted at 20 mg/mL in dichloromethane and sodium trifluoroacetate (Na + TFA − ) diluted at 5 mg/mL in acetone for P(FDA) homopolymer, and P(AAEM-co-FDA), P(StySAc-co-FDA) and P(4VP-co-FDA) gradient copolymers. The polymer concentration was 20 mg/mL in trifluorotoluene (C 6 H 5 CF 3 ). The polymer and matrix were mixed in a 1:1 volume ratio. The Na + TFA − layer was first deposited on the target and dried. After evaporation of solvent, the mixture, composed of polymer and matrix, was placed on the top of the sodium trifluoroacetate layer in the MALDI target.
Regarding the P(DPPS-co-FDA) gradient copolymer, the matrix was dithranol diluted at 20 mg/mL in dichloromethane and the cationizing agent was NaI diluted at 5 mg/mL in acetone. The polymer concentration was 10 mg/mL in hexafluoro-isopropanol (HFIP). The polymer and matrix were mixed in a 1:5 volume ratio. The NaI layer was first deposited on the target and dried. After evaporation of solvent, the mixture, composed of polymer and matrix, was placed on the top of the NaI layer in the MALDI target.

Synthesis and Characterization of the (Co)Polymers before Aminolysis
The synthesis of the proposed (co)polymers ( Figure 1) has been achieved through the reversible-deactivation radical polymerization technique, specifically reversible additionfragmentation chain-transfer (RAFT) polymerization [66]. RAFT polymerization is compatible with many functional groups. This technique allows preparing gradient copolymers, thanks to the reactivity ratio of the different monomers. Indeed, the Alfrey and Price Q and e values of FDA, 4VP and AAEM are available in the literature: Q FDA = 0.44 and e FDA = 0.45 [74], Q 4VP = 2.47 and e 4VP = 0.84, [75], Q AAEM = 0.68 and e AAEM = 0.13 [76]. The Q and e values of chloromethyl styrene (CMSty) have been considered for DPPS and StySAc as a first approximation: Q CMSty = 1.13 and e CMSty = −0.58 [77]. From these Q and e values, the reactivity ratios could be estimated by using the classical Alfrey and Price equation [75]: r 4VP = 4.05 and r FDA = 0.21, r AAEM = 1.61 and r FDA = 0.56, r CMSty = 1.41 and r FDA = 0.24. Thus, the complexing monomers are preferably introduced during copolymerization with FDA, (i.e., polymer chains are first enriched with complexing units), leading to gradient copolymer structures. In addition, we observed by 1 H-NMR of the crude reaction medium at a given time that the complexing monomers (4VP, AAEM, DPPS, StySAc) polymerize faster than FDA, which confirmed the theoretical data ( Figures S5-S7). Thus, the beginning of the polymer chain is enriched in the complexing monomer units whereas the end of the polymer chain is enriched in the FDA CO 2 -philic monomer units (gradient structure).
The targeted molecular weight (M n,targeted ) (Equation (S1)) was set to 5000 g/mol for P(FDA) and 10,000 g/mol for the other copolymers. The AIBN/CTA molar ratio was set to 0.3 in all cases and polymerizations were performed at 65 • C in TFT or toluene in the case of P(4VP-co-FDA). The associated results are gathered in Table 1. P(FDA) with a molecular weight of around 5000 g/mol and four copolymers with a molecular weight of ca. 10,000 g/mol were successfully obtained. All the polymers were purified by precipitation in pentane. P(FDA) homopolymer and the respective copolymers with DPPS and 4VP were obtained as a fine red powder. Instead, the fluorinated copolymers containing AAEM and StySAc units were isolated as a pink sticky gum after precipitation. After drying under a vacuum, they become fragile foams that can be easily milled into powder. High monomer conversions were reached (from 80 to 100% conversion) and quite good correlations are observed between M n,targeted and M n,NMR . The higher values obtained for M n,NMR may be explained by (co)polymer chain fractionation during the precipitation step, causing the loss of lower molecular weight chains. 1 H-NMR analyses were performed to determine polymer chain lengths, as well as the ratio between complexing and CO 2 -philic monomer units (SI Section S2). These analyses have been performed using acetone-d 6 for AAEM copolymer [78], CDCl 3 for 4VP and StySAc copolymers, and CFC-113 as a solvent with C 6 D 6 capillaries (for locking) for the other (co)polymers. The selection of this unusual solvent was due to the different solubility of the polymers in common solvents: the fluorinated copolymers with AAEM and StySAc units were also soluble in acetone and ethyl acetate, while the copolymer with 4VP is soluble in chloroform; contrarily, the other FDA homo-and copolymers were soluble exclusively in fluorinated solvents such as TFT or CFC-113. To calculate the polymer composition, the aromatic signal of the RAFT agent has been used as a reference. 1 H-NMR of the P(FDA) homopolymer permits to easily identify the typical signal of the fluorinated units once polymerized, as well as the CTA signals ( Figure 2). This information is useful for the more complicated copolymer NMR analyses. The signal of the aromatic protons of the CTA is visible between δ = 7.5 and 8.3 ppm. The signal at δ = 5 ppm corresponds to the proton of the last FDA monomer unit, in α-position to the thioester group of the CTA chain end. Another typical signal is present at δ = 4.5 ppm, which is assigned to the methylene protons of the fluorinated alkyl group and is used to calculate the number of FDA monomer units in the polymer. To calculate the mean number and average degree of polymerization DP FDA and the molecular weight M n,precipitated,P(FDA) for the P(FDA), Equations (S4) and (S5) were used. M n,precipitated,P(FDA) equal to 5850 g/mol and DP FDA of 10.8 were estimated by 1 H-NMR. 1 H-NMR spectrum of P(4VP-co-FDA) showed the appearance of two broad peaks at δ = 6.5 and 8.5 ppm, in addition to the typical FDA signal at δ = 4 ppm (Figure 3). These signals are assigned to the pyridinic protons. As for the P(FDA) homopolymer, Equations (S8)-(S10) can be used to determine the degrees of polymerization and the molecular weight of the copolymer. A copolymer with a molecular weight of 11,820 g/mol and degrees of polymerization of 19.8 for the 4VP complexing monomer and 18.3 for FDA was obtained.
The 1 H-NMR spectrum obtained after FDA and AAEM copolymerization and recorded using acetone-d 6 as a solvent, shows two different forms of the AAEM moiety, as ketone and as enol, with 4.5 mol% of the enolic conformation ( Figure 4).
The calculation of the degrees of polymerization and polymer molecular weight has been performed by taking into account both ketone and enol forms of the AAEM monomer units, using Equations (S13)-(S15). We obtained a molecular weight equal to 13,480 g/mol with DP AAEM = 18.9 and DP FDA = 17.7 for the copolymer P(AAEM-co-FDA) according to 1 H-NMR spectrum ( Figure 4). P(DPPS-co-FDA) copolymer is only soluble in TFT and CFC-113 solvents, allowing exclusively to perform the 1 H-NMR analysis using C 6 D 6 capillaries for locking ( Figure 5). A broad peak between δ = 6.5 and 8 ppm can be observed, related to the aromatic protons of the phosphonic monomer.  The degrees of polymerization and the molecular weight of the copolymer P(DPPSco-FDA) were estimated using Equations (S19)-(S21). They were estimated to be equal to M n,precipitated,P(DPPS-co-FDA) = 11,640 g/mol with DP DPPS = 7.5 and DP FDA = 17.8 ( Figure 5). P(StySAc-co-FDA) copolymer is slightly soluble in acetone, and soluble in chloroform. Its 1 H-NMR spectrum shows a broad peak between δ = 6.5 and 7 ppm relative to the aromatic protons of the StySAc monomer ( Figure 6). To calculate the degrees of polymerization and the molecular weight of the P(StySAc-co-FDA), it needs to take into account the overlapping at δ = 4 ppm of the typical methylene signal of the FDA units with the benzylic signal of StySAc units, using Equations (S26)-(S28). A copolymer with a molecular weight of 9790 g/mol and degrees of polymerization of 9.7 for the StySAc complexing monomer and 14.8 for FDA was obtained.
Through all these data, it was possible to estimate the weight fractions of complexing monomer units and FDA monomer units for all the copolymers according to general Equations (S2) and (S3), respectively, (see SI Section S2). Weight fractions of complexing monomer units ranging from 0wt% for P(FDA) homopolymer to 30.6wt% for P(AAEM-co-FDA) copolymer were thus estimated.
(Co)polymer structures were also confirmed by MALDI-TOF-MS analyses using reflectron positive ion mode (see SI Section S2.6).
The MALDI-TOF-MS mass spectrum of P(AAEM-co-FDA) was detected between 1800 and 5750 m/z (Figure 7) although M n,NMR of the precipitated copolymer was estimated to be equal to 13,480 g/mol. Such discrepancy is not surprising as it is well known that high-accuracy mass measurements are possible for oligomer samples only with a mass range no greater than 3000 m/z [79].
The main population corresponds to the species ionized with sodium with the formula CH 3 CH 2 OC(O)CHCH 3 (AAEM n -co-FDA m )SC(S)C 6 H 5 + Na + , i.e., one CTA end-group associated with some units of AAEM and some units of FDA monomer incremented with one sodium cation due to ionization.
A more detailed analysis of the MALDI-TOF-MS data in the region corresponding to the strongest peaks (2500-3610 m/z; Figure 8 Table 2. Each peak of the copolymer was calculated according to Equation (1)     Lastly, note that the isotope patterns of theoretical and experimental sodium adducts of P(AAEM-co-FDA) copolymer exhibit a similar shape and are therefore in good agreement ( Figure 9). Hence, the MALDI-TOF spectrum confirms the formation of the P(AAEM-co-FDA) copolymer.

Aminolysis of the (Co)Polymers
We took advantage of the RAFT technique to introduce a thiol group in our (co)polymers by aminolysis of the (co)polymer chain end. Indeed, thiols are known to complex some metals, (e.g., gold) and can therefore bring additional properties to our CO 2 -soluble metalcomplexing (co)polymers. The aminolysis of the different fluorinated (co)polymers has been performed using three equivalents of piperidine. In order to avoid disulfide formation, triphenylphosphine was added as a reductant in a previously degassed solution of the polymer in TFT (SI Section S3).
Usually, end-group removal of RAFT polymers is performed by primary amines, such as hexylamine, butylamine, or cyclohexylamine [80,81]. The reaction can be easily followed through the color change of the polymer solution upon aminolysis. Indeed, a few seconds after the addition of piperidine, the solution switches from red color (typical of the phenylcarbonothioyl terminal group derived by the CTA) to a bright yellow color (typical of the phenylthioamide, which is the by-product of the aminolysis reaction) as shown in Figure 10. Even if this end-group removal method can be applied to all our synthesized polymers, the use of n-butylamine (primary amine) led to a secondary reaction for the fluorinated copolymer containing AAEM units. Indeed, comparing the 1 H-NMR spectra of the polymer before and after aminolysis by n-butylamine, it is possible to see some differences in the signal related to the acetoacetoxy protons ( Figure 11).
In Figure 11, it is possible to identify two main peaks. The first peak (δ = 4-4.5 ppm) is related to the ethylenic protons of each monomer unit; meanwhile, the second (δ = 3.7 ppm) is attributed to the acetoacetate protons. The ratio between them changes depending on the different amines used to perform the aminolysis. This change is most probably due to the formation of enamine between butylamine and the acetoacetate group of AAEM [82,83]. Thus, for the sake of comparison, all the aminolyzed polymers reported in this work were obtained by using piperidine (secondary amine) as nucleophile. The 1 H-NMR spectra of the aminolyzed copolymers are given in the SI (Figures S19-S24).

Thermal Polymers Characterization Pre-and Post-Aminolysis
The polymers before and after aminolysis were thermally characterized in order to investigate their thermal stability (TGA analyses under nitrogen) as well as the presence of thermal transitions (DSC analysis) (SI, Figures S25-S44) (Table 3). These analyses have been performed to study the influence of the different metal-complexing units of the copolymers and chain ends (pre-and post-aminolysis) on such gradient copolymers. Table 3. Thermal properties of the synthesized polymers. TGA analyses show T 2wt%loss (temperature where 2% of the weight is lost) higher than 200 • C for all the polymers, except the aminolyzed P(FDA)SH homopolymer (197 • C) and P(StySAc-co-FDA) copolymer (160 • C). In addition, higher T 2wt%loss are obtained for the copolymers after aminolysis (except for P(FDA) homopolymer): this behavior can be explained by the thermal elimination of the terminal phenyldithio group. Indeed, the cleavage of the dithio group in the case of unreacted CTA (R-S-C(S)Z, in this work, R = EtOC(O)CH(CH 3 )-and Z = Ph) occurs close to 200 • C, depending on the R group of the CTA. The higher residual mass by TGA after removing the end-group coming from the RAFT agent is also consistent with some works showing the lower thermal stability of RAFT polymers, even sometimes opening the route to a free depolymerization approach [84,85]. On the other hand, it is important to mention that the stability of the R group is increased by the integration of the CTA inside the polymer chain [81,86,87]. Thus, the following order of thermal stability was determined according to TGA results: P(FDA) > P(4VP-co-FDA) > P(AAEM-co-FDA) > P(DPPS-co-FDA) > P(StySAc-co-FDA).
DSC analyses show that P(FDA) and P(FDA)SH homopolymers present a melting temperature (T m ), respectively, at 63 and 56 • C, before and after aminolysis, while no glass transition (T g ) was detected for both homopolymers. On the other hand, all the gradient copolymers exhibit a glass transition temperature and a melting temperature before and after aminolysis, except P(DPPS-co-FDA) and P(DPPS-co-FDA)SH copolymers that are 100% amorphous with only a T g visible at 67 and 68 • C, respectively. Note that T g -values are higher after aminolysis of the gradient copolymers. A decrease in the T g suggests lower intermolecular interactions between the polymer chains, which can be caused by the lower Lewis base character of AAEM and StySAc moieties compared to 4VP and DPPS.

Polymers Phase Behavior in Dense CO 2
The phase behavior in dense CO 2 was studied for all the synthesized polymers, in both their pre-and post-aminolysis forms (SI Section S5, Figures S45-S55). To this aim, the cloud point measurements were performed in liquid and supercritical CO 2 . This analysis allows determining the transition from a cloudy CO 2 -polymer suspension to a transparent solution. This transition indicates the cloud point from the solubilized polymer (transparent solution) to the cloudy suspension (polymer insoluble) and vice versa. The obtained data, recorded in a cooling process (meaning screening the temperature from 65 down to 25 • C), are shown in Figure 12 and Figure S55.
All the synthesized polymers are soluble at pressures lower than 27 MPa up to 65 • C, which can be considered mild conditions. For the sake of comparison, decaffeination by scCO 2 , which is a major industrial application of supercritical fluids, is routinely operated at a pressure of 22 MPa and a temperature of 90 • C [88]. Mild operating conditions represent an important parameter for the intended application of the synthesized polymers in metal recovery. Notably, three different trends can be identified in Figure 12 from the non-aminolyzed polymers. The first one corresponds to the P(FDA) homopolymer which shows the lowest cloud point pressures at a given temperature (Figure 12a). A second trend corresponds to the fluorinated copolymers incorporating AAEM and StySAc units, showing intermediate cloud point pressures (Figure 12b). Finally, a third trend corresponds to the copolymer containing DPPS and 4VP units, showing the highest cloud point pressures (Figure 12c). Regarding these last two different trends, it is interesting to notice that, going back to the purification step of the copolymers, the ones with lower cloud point pressures formed a gum during the precipitation. Furthermore, AAEM and StySAc copolymers showed a rather low T g (8 and 24 • C, respectively). On the contrary, the copolymer containing DPPS and 4VP units have shown an easier precipitation behavior during the purification step, forming a powder and showing a higher glass transition temperature (67 and 61 • C, respectively). These facts allow us to draw a correlation between the T g of the copolymers and their solubility in dense CO 2 : the higher T g indicates an increased interaction between the polymer chains, decreasing the solubility of the polymers in dense CO 2 [89]. These behaviors can be additionally explained by the nature of the comonomers. Indeed, it has been proved that monomers such as pyridine have a favorable nitrogen-nitrogen interaction instead of nitrogen-CO 2 once incorporated in a polymeric chain, bringing a reduced solubility in dense CO 2 [90]. Instead, the addition of acetate groups in the polymer chain increases the number of electron-rich groups in the system, improving its CO 2 -philicity [91]. In summary, the following decreasing order of solubility of the polymers has been determined from the cloud point curves: P(FDA) > P(StySAc-co-FDA) ∼ = P(AAEM-co-FDA) > P(DPPS-co-FDA) ∼ = P(4VP-co-FDA).
Furthermore, the solubility of the aminolyzed polymers has also been investigated in dense CO 2 . Interestingly, the aminolysis reaction, with the introduction of a polar thiol group as polymer chain end instead of dithiobenzoate moiety, did not influence significantly the solubility of these polymers: the cloud point pressure only increased by a few MPa (less than 3.2 MPa, in the worst case of P(AAEM-co-FDA) at 65 • C) at a given temperature. Thus, this limited difference in cloud point pressure does not preclude the good solubility in dense CO 2 of such thiol-terminated polymers. The mild conditions necessary to solubilize this wide range of polymers in a green solvent such as CO 2 demonstrate the remarkable relevance of this library of metal-complexing CO 2 -soluble polymers for their application as extractants for the removal of metals from industrial or urban waste.

Ability of Polymers to Complex Metals
Preliminary experiments have already been performed by our team to verify the ability of such polymers to complex metals. Thus, P(AAEM-co-FDA) copolymers showed their ability to be used as palladium ligands [92]. Complexation of palladium acetate was performed by the complexing AAEM units of the copolymer leading to a P(AAEM-co-FDA)-Pd II supramolecular complex. The resulting polymer-supported Pd catalyst was successfully used for various applications such as a catalyst for Heck reaction, generation of Pd 0 nanoparticles in the polymer matrix, or Pd II impregnation of mesoporous silica in scCO 2 followed by reduction to Pd 0 (preparation of silica-supported Pd catalyst). In addition to Pd, P(AAEM-co-FDA) copolymer has also shown its ability to complex cobalt and more precisely cobalt acetate hydrate. Complexes Co-P(AAEM-co-FDA) exhibit cloud point pressure approximately 30 to 40 bar higher than the copolymer alone [93]. Thiolterminated homopolymer P(FDA)SH, P(DPPS-co-FDA) and aminolized P(DPPS-co-FDA)SH gradient copolymers were used to extract palladium from commercial Pd/Al 2 O 3 supported catalysts [53]. Extractions were performed in scCO 2 under mild conditions (40 • C and 25 MPa). Polymers were found to successfully extract up to ca. 40% of Pd from the Pd/Al 2 O 3 catalysts. P(DPPS-co-FDA)SH was the most efficient system, requiring a much lower quantity than the two other polymers to achieve the same level of extraction in one hour. Lastly, P(4VP-co-FDA) and aminolyzed P(4VP-co-FDA)SH gradient copolymers showed their ability to extract Pd from the aluminosilica-supported catalyst in scCO 2 under mild conditions (40 • C and 25 MPa). More precisely, ca. 70% of the Pd was removed from an aluminosilica support in the presence of P(4VP-co-FDA) copolymer [54].
These different metal-complexing CO 2 -philic polymers demonstrated their ability to be used for various applications. These promising results urged us to enlarge the panel of such polymers and to detail their synthesis method.

Conclusions
In this work, we have presented the synthesis of ten (co)polymers which are constituted by a CO 2 -philic fraction, represented by FDA monomer units, and by CO 2 -phobic complexing monomer units (4VP, AAEM, DPPS, and StySAc). These (co)polymers were successfully synthesized by the reversible-deactivation radical polymerization technique (RAFT). 1 H-NMR characterization allowed us to determine the composition as well as the molecular weight of the synthesized (co)polymers. In addition, the thermal properties of the (co)polymers have been studied by TGA and DSC, which confirms the good thermal stability of the (co)polymers (decreasing order of thermal stability: T 2wt%loss,P(FDA) = 263 • C > T 2wt%loss,P(4VP-co-FDA) = 249 • C > T 2wt%loss,P(AAEM-co-FDA) = 239 • C > T 2wt%loss,P(DPPS-co-FDA) = 215 • C > T 2wt%loss,P(StySAc-co-FDA) = 160 • C) and their difference in T m (T m,P(AAEM-co-FDA) = 41 • C < T m,P(StySAc-co-FDA) = 51 • C < T m,P(FDA) = 63 • C < T m,P(4VP-co-FDA) = 85 • C) and T g transition temperatures (T g,P(AAEM-co-FDA) = 8 • C < T g,P(StySAc-co-FDA) = 24 • C < T g,P(4VP-co-FDA) = 61 • C < T g,P(DPPS-co-FDA) = 67 • C). Finally, their phase behavior in dense CO 2 has been studied by cloud point measurements, showing that these metal-complexing (co)polymers are soluble at mild pressure conditions (lower than 27 MPa at 65 • C), which is a key factor for their eventual application as metal extraction agents. Through the synthesis of well-defined copolymers (gradient architecture, monomer composition, and polymer molecular weight) and their promising CO 2 solubility, this library of metal-complexing CO 2 -soluble polymers show good candidates for the recovery or the decontamination of end-of-life metals-containing liquid or solid wastes. The investigation of their performances in metal extractions from supported catalysts is in progress and will be reported in forthcoming articles.

Data Availability Statement:
The data presented in this study are available in the Supporting Information.

Conflicts of Interest:
The authors declare no conflict of interest.