How to easily adapt cloud points of statistical thermosensitive polyacrylamide-based copolymers knowing reactivity ratios

Abstract

The present contribution deals with the ease of controlling the thermosensitivity of statistical copolymers prepared by free radical polymerization playing on the reactivity ratios of both monomers used for the synthesis. The copolymers were prepared using N-n-propylacrylamide (NnPAAm) monomer to achieve the thermoresponsive properties. The second monomer was either (dimethoxyphosphoryl)methyl 2-methylacrylate (MAPC1) or diethyl 2-(acrylamido)ethylphosphonate (DAAmEP) in order to incorporate phosphonic acid moieties after hydrolysis. The architecture of these original statistical copolymers was determined by measuring the reactivity ratios of both (NnPAAm/MAPC1) and (NnPAAm/DAAmEP) monomer couples. By varying the chemical nature of the phosphonated-based monomer (methacrylate or acrylamide), it was possible to get different reactivity ratios, and as a consequence to significantly affect the thermosensitive behavior of the statistical copolymers for a constant proportion of phosphonic acid groups. Additionally, all copolymers showed similar sorption rates toward nickel cations (Ni²⁺).

---

Introduction

Thermosensitive polymers have been widely studied in recent decades as they show interesting behavior as changes of physical properties are triggered by temperature stimulus.^1,2 As a result, such polymers can be potentially useful for many applications, in particular in the biomedical field,^1,3–5 surface coatings^6,7 or water treatment.^8–11 Such polymers are soluble in water at low temperature and become non-soluble upon increasing the temperature. The change of physical state occurs at a temperature called the cloud point (CP), often considered as the Lower Critical Solution Temperature (LCST) by approximation. The change of solubility is linked to the competition between polymer–water and polymer–polymer interactions. Below the LCST, the presence of hydrogen bonds between the polymer and the water as well as the formation of solvation cages explains the solubility of the polymer.¹² Increasing the temperature above the LCST causes the hydrogen bonds to break, favors the polymer–polymer interactions and leads to the loss of the solubility.

Such an evolution of solubility has been described for a wide range of polymers such as poly(vinyl methyl ether),^13,14 poly(N-vinylcaprolactam),^15–17 poly(ethylene oxide)^18,19 or polyacrylamides,^1,20 enabling to reach various LCST values ranging from −5 °C to higher than 100 °C. LCST are directly linked to the chemical structure of the polymer and, for a given polymer, can be modified by the incorporation of a comonomer into the polymer chains. Indeed, the LCST increases when the comonomer is hydrophilic as the interaction of the resulting copolymer with water becomes stronger than for the corresponding homopolymer. On the contrary, the addition of hydrophobic monomers leads to weaker copolymer–water interactions and thus to lower LCST values.¹ So, it is possible to regulate copolymer LCST values by the incorporation, in appropriate proportions, of well chosen comonomers.

Another way of varying the LCST consists of playing with the architecture of the copolymers.¹ To date, examples reported in the literature have only dealt with the comparison of different kinds of well-defined copolymers. For instance, the effect of poly(oxyethylene) (PEO) on the thermal response of aqueous solutions of poly(N-isopropylacrylamide) (PNIPAM) block and graft copolymers was studied.²¹ PNIPAM-b-PEO showed thermoresponsive properties similar to the PNIPAM homopolymer. On the contrary, aqueous PNIPAM-g-PEO showed a decrease in demixing temperature when incorporating 6–10 grafts of PEO, indicating the influence of the macromolecular architecture. Additionally, the thermal response of the aqueous PNIPAM-g-PEO was significantly faster than in the case of either pure PNIPAM or PNIPAM-b-PEO. A different contribution described the synthesis of block copolymers made of methoxy diethylene glycol acrylate and short hydrophobic polystyrene end blocks (B blocks) by means of RAFT polymerization. Three different architectures, namely amphiphilic diblock AB, symmetrical triblock BAB, and three-arm star (BA)₃ were prepared using three different chain transfer agents. The CPs of the various polymers proved to be sensitive to the architecture, higher CPs being obtained for the diblock copolymers.²² Another example dealt with the comparison between linear and hyperbranched thermo-responsive copolymers based on poly(ethylene glycol) methacrylates.²³ Materials were synthesized by reversible addition fragmentation chain transfer polymerization (RAFT) of di(ethylene glycol) methacrylate and oligo(ethylene glycol) methacrylate with ethylene glycol dimethacrylate as crosslinker. The thermoresponsive properties of the hyperbranched copolymers were compared with their linear analogs. Molecular architecture influenced thermosensitive behavior, with a decrease of around 5–10 °C in the LCST of the hyperbranched polymers compared with the LCST of the corresponding linear chains. Finally, to our knowledge, only one example was described in the literature involving differences between statistical and block copolymers.²⁴ ABA-type triblock copolymers with n-butyl methacrylate and 2-(dimethylamino)ethyl methacrylate (DMAEMA) as the A and the B blocks, respectively, were synthesized. These triblock copolymers did not show a cloud point, whereas the corresponding statistical copolymers did. A possible explanation was that the triblock copolymers formed micelles and stabilized themselves in solutions while the statistical copolymers were not able to form micelles, precipitating out.

From the results described in the literature, there is no doubt that the architecture has a non-negligible impact on the LCST value. Until now, the only way used to vary and thus adapt the LCST of a particular copolymer was to achieve the synthesis of well-defined structures (block, graft, star), using a non versatile polymerization process. In the present contribution, we describe a way to adapt the LCST value of statistical copolymers prepared by free radical polymerization, changing the chemical nature of the comonomer. For such purposes, the synthesis of original thermoresponsive copolymers based on poly(N-n-propylacrylamide) (NnPAAm) and bearing phosphonic acids moieties was achieved. The incorporation of phosphonated groups was obtained by copolymerization of the NnPAAm monomer with two phosphonated comonomers, (dimethoxyphosphoryl) methyl 2-methylacrylate (MAPC1) and diethyl 2-(acrylamido)ethylphosphonate (DAAmEP), i.e. a methacrylate and an acrylamide monomer, respectively. Three different copolymers have been synthesized. Two of them, P(NnPAAm-stat-MAPC1) and P(NnPAAm-stat-DAAmEP), have been obtained by free radical copolymerization of (i) NnPAAm and MAPC1 and (ii) NnPAAm and DAAmEP monomers, respectively. The final copolymer, P(NnPAAm-block-DAAmEP), was prepared by a two step RAFT polymerization process²⁵ and was used to compare both statistical and block chemical structures. Theoretical architecture of statistical copolymers was determined by measuring the reactivity ratios of monomer couples, i.e. NnPAAm/MAPC1 and NnPAAm/DAAmEP. For all copolymers, an additional hydrolysis step of the phosphonated ester into phosphonic acid groups was achieved leading to P(NnPAAm-stat-_hMAPC1) and P(NnPAAm-stat-AAmEPA) statistical copolymers and P(NnPAAm-block-AAmEPA) diblock copolymer. Once phosphonic acid groups were obtained, it appeared interesting to study the influence of the copolymer architectures on the sorption properties of nickel cations (Ni²⁺) in order to determine if it was possible to reach high sorption properties while keeping low cloud points values, which is notably a key parameter for water treatment applications. The ideal system would result in combining “control” of the CP as a function of the monomer nature with using a versatile polymerization technique such as radical polymerization.

Experimental

Materials

N-n-propylacrylamide (NnPAAm, [25999-13-7] – SP43-0-002) and (dimethoxyphosphoryl) methyl 2-methylacrylate (MAPC1, [86242-61-7] – SP41-003) were supplied by Specific Polymers (Montpellier, France) and used as received. Dimethylsulfoxide (DMSO), and 1,4-dioxane were supplied from Sigma-Aldrich and used as received. N-2-(Bromoethyl)phtalimide (Alfa Aesar, 98%), hydrazine monohydrate (Alfa Aesar, 98%), triethylphosphite (Aldrich, 98%), poly(4-vinylpyridine) (Aldrich, 2% cross-linked), acryloyl chloride (Aldrich, 97%), 2-cyano-2-propyl dodecyl trithiocarbonate (CTA) (Aldrich, 97%), and bromotrimethylsilane (Aldrich, 97%) were used as received. Azobisisobutyronitrile (AIBN) (Aldrich, 98%) was used after recrystallization in methanol.

Analytical techniques

¹H and ³¹P NMR spectra were recorded using a Bruker Avance DRX 200 (200 MHz) with CDCl₃ or D₂O as deuterated solvents. For ¹H NMR, chemical shifts were referenced to the peak of residual non-deuterated solvents at 7.3 ppm and 4.8 ppm for CDCl₃ and D₂O, respectively. 2D DOSY NMR experiments were performed on a Bruker Avance 3 (600 MHz) at 25 °C in DMSO-d₆ as solvent. Size exclusion chromatography in N,N-dimethylacetamide (DMAc) was performed on a PL-GPC 50 Plus equipped with a RI refractive index detector. PolarGel M columns were used at 50 °C with a 0.8 mL × min⁻¹ flow rate of DMAc (+0.1% LiCl weight), calibration was achieved using polystyrene (PS) standards. The thermosensitivity of the polymers was estimated by a change in the transmittance through polymer aqueous solutions during a gradual increase of the temperature. The measurement of the transmittance was carried out on copolymer aqueous solutions (5 g L⁻¹) with a Perkin Elmer Lambda 35 UV-Visible spectrometer equipped with a Peltier temperature programmer PTP-1+1. A wavelength of 500 nm was selected. The temperature ramp was 0.2 °C min⁻¹ between 15 °C and 40 °C. The thermosensitivity was determined at the sudden slope change in the transmittance curve. The LCST values of the copolymers thus corresponded to the minimum of the derivative curves. Sorption experiments were carried out as already described in a previous paper.⁹ The typical procedure is recapped in the ESI.† The amount of Ni²⁺ ions trapped by the copolymers was determined by measuring the concentration of Ni cations in the solution before and after sorption experiments. Metal ion concentrations were assessed by atomic adsorption spectroscopy with a Perkin Elmer AAnalyst 400, an AutoPrep 50 dilutor and a S10 Auto-sampler. Experiments were carried out over 24 hours to reach equilibrium and the temperature was controlled by a 6 Liter Fisher Bioblock Scientific Cryothermostated bath. The influence of the molar ratio between Ni²⁺ metal cations and phosphonated moieties was studied by changing the amount of copolymers involved in the sorption experiments.

Synthesis of diethyl 2-(acrylamido)ethylphosphonate monomer (DAAmEP)

The synthesis of diethyl 2-(acrylamido)ethylphosphonate monomer (DAAmEP) was described in our previous paper.²⁶

Synthesis of P(NnPAAm-stat-MAPC1) copolymers

All P(NnPAAm-stat-_hMAPC1) syntheses were carried out in dioxane at 70 °C in Schlenk tubes by free radical polymerization using AIBN as radical initiator. A typical procedure for the synthesis of poly(NnPAAm-stat-_hMAPC1) 90/10 is described here: both monomers NnPAAm (3.0 g, 2.65 × 10⁻² mol) and MAPC1 (0.62 g, 2.94 × 10⁻³ mol), dioxane (20 mL) and AIBN (24.2 mg, 1.47 × 10⁻⁴ mol) were first introduced into the Schlenk tube and the obtained solution was degassed by three freeze–pump–thaw cycles and then heated at 70 °C. Polymerizations lasted 8 hours. The obtained copolymers were purified by precipitation in a large volume of cold hexane and dried under reduced pressure (M_n = 55 700 g mol⁻¹, dispersity (Đ) = 3.3).

¹H NMR (D₂O, 200 MHz, δ): 4.44–3.97 (OCH^′₂), 3.97–3.92 (P(O)(OCH^′₃)₂), 3.25–2.76 (CH₂N), 2.47–1.86 (CH₂–CH), 1.86–1.24 (CH₂–CH, CH^′₂CH′, CH₂–CH₂–N), 1.22–0.98 (C–CH^′₃), 0.98–0.71 (CH₃).

³¹P NMR (D₂O, 81 MHz, δ): 24.0 (1P, P(O)(OCH^′₃)₂).

Synthesis of P(NnPAAm-stat-DAAmEP) copolymers

All P(NnPAAm-stat-AAmEPA) syntheses were carried out in dioxane at 70 °C in Schlenk tubes by free radical polymerization using AIBN as radical initiator. A typical procedure for the synthesis of poly(NnPAAm-stat-AAmEPA) 90/10 is described here: both monomers NnPAAm (3.0 g, 2.65 × 10⁻² mol) and DAAmEP (0.69 g, 2.94 × 10⁻³ mol), dioxane (10 mL) and AIBN (24.2 mg, 1.47 × 10⁻⁴ mol) were first introduced into a Schlenk tube and the obtained solution was degassed by three freeze–pump–thaw cycles and then heated at 70 °C. Polymerizations lasted 8 hours. The obtained copolymers were purified by precipitation in a large volume of cold hexane and dried under reduced pressure (M_n = 86 000 g mol⁻¹, Đ = 3.6).

¹H NMR (D₂O, 200 MHz, δ): 4.31–4.06 (O–CH^′₂–CH^′₃), 3.57–3.31 (CH^′₂–N), 3.29–2.89 (CH₂–N), 2.35–1.90 (CH₂–CH₂–P), 1.90–1.44 (CH₂–CH, CH^′₂CH′, CH^′₂–CH^′₂–P), 1.25–1.06 (O–CH^′₂–CH^′₃), 1.06–0.81 (CH₂–CH₃).

³¹P NMR (D₂O, 81 MHz, δ): 31.2 (1P, P(O)(O–CH^′₂–CH^′₃)₂).

Synthesis of P(NnPAAm-block-DAAmEP) copolymers

The first step was the synthesis of poly(N-n-propylacrylamide) macro-chain transfer agent by reversible addition fragmentation transfer (RAFT) polymerization. A typical procedure is described here for the synthesis of poly(N-n-propylacrylamide) using 2-cyano-2-propyl dodecyl trithiocarbonate (CTA) as chain transfer agent for a targeted molecular weight of 10 000 g mol⁻¹; [CTA]/[AIBN] ratio was 4 : 1. CTA (110.45 mg, 3.2 × 10⁻⁴ mol), AIBN (13.5 mg, 8.0 × 10⁻⁵ mmol) and NnPAAm (4.0 g, 3.5 × 10⁻² mol) were added along with 15 mL of DMSO to a 30 mL Schlenk tube. The mixture was degassed by four freeze–pump–thaw cycles and then heated at 70 °C under nitrogen in a thermostat-controlled oil bath. The conversion was measured by ¹H NMR in deuterated chloroform comparing the signals of the reactive double bond (5.6 and 6.1 ppm) to the signals of the methylene in α-position of amide group at 2.8–3.1 ppm. The polymerization proceeded for 130 min before the P(NnPAAm) product was purified by precipitation in distilled water at 45 °C and dried under reduced pressure (conversion = 76%, M_n = 14 800 g mol⁻¹, Đ = 1.27).

¹HNMR (CDCl₃, 200 MHz, δ): 3.30–2.90 (CH₂N), 2.25–1.85 (CH₂–CH), 1.85–1.30 (CH₂CH₂NH and CH₂–CH), 1.00–0.70 (CH₃).

The second step corresponded to the synthesis of the P(NnPAAm-block-DAAmEP) copolymer by reversible addition fragmentation transfer (RAFT) polymerization using the P(NnPAAm) previously synthesized as a macro-chain transfer agent. A typical procedure for the synthesis of the P(NnPAAm-block-DAAmEP) block copolymer is described here for a targeted molecular weight of 5000 g mol⁻¹ for the DAAmEP block: DAAmEP (0.5 g, 2.12 × 10⁻³ mol), P(NnPAAm) (0.8 g, 8.0 × 10⁻⁵ mol) and AIBN (6.5 mg, 3.95 × 10⁻⁵ mol) were dissolved in 10 mL of DMSO in a 30 mL Schlenk tube. A chemically inert compound, 1,3,5-trioxane, was added to the reaction mixture in order to determine the conversion as time proceeded for the polymerization. The mixture was degassed by three freeze–pump–thaw cycles and then heated at 70 °C under nitrogen in a thermostat-controlled oil bath. The polymerization proceeded for 210 min (corresponding to a conversion of 79%) before P(NnPAAm-block-DAAmEP) was purified. The DMSO was first eliminated by cryo-distillation. Crude P(NnPAAm-block-DAAmEP) was then dissolved in acetone, precipitated in cold hexane and dried under reduced pressure (conversion = 83%, M_n = 19 400 g mol⁻¹, Đ = 1.30).

¹H NMR (CDCl₃, 200 MHz, δ): 4.40–3.95 (OCH^′₂), 3.90–3.40 (CH^′₂N), 3.35–2.95 (CH₂N), 2.5–1.80 (CH₂–CH and CH^′₂–CH′), 1.80–1.45 (CH₂CH₂N, CH₂–CH, CH^′₂CH^′₂N and CH^′₂–CH′), 1.45–1.10 (CH^′₃), 1.05–0.80 (CH₃).

³¹P NMR (CDCl₃, 81 MHz, δ): 29.3 (1P, P(O)(OEt)₂).

Hydrolysis of the phophonated ester groups into phosphonic acid groups

For all obtained copolymers, an additional synthesis step was performed in order to hydrolyze the phosphonated esters into phosphonic acid groups. The hydrolysis of dimethyl or diethylphosphoryl (O=P–(OR)₂, R = Me, Et) functions was achieved using bromotrimethylsilane (BrSiMe₃) and methanol as already described in the literature.^27–29 Copolymers were finally washed by dialysis against pure methanol in order to remove unreacted molecules. P(NnPAAm-stat-_hMAPC1), P(NnPAAm-stat-AAmEPA) and P(NnPAAm-block-AAmEPA) were obtained from P(NnPAAm-stat-MAPC1), P(NnPAAm-stat-DAAmEP) and P(NnPAAm-block-DAAmEP), respectively.

Theory dealing with the statistical copolymer composition

By free radical copolymerization, the composition of the copolymer chains usually differs from the composition of the comonomer feed. Indeed, during the propagation reactions, two monomers M₁ and M₂ can be added either to a propagating chain ending by M₁ or by M₂ as shown in eqn (1)–(4).³⁰where k_ij is the rate constant for the propagation of an M_j monomer to a chain ending with an M_i monomer. Copolymer chains compositions are then characterized by two parameters, which only depend on the monomers respective reactivity: the reactivity ratios r. These parameters are defined as ratio of the rate constant for a reactive propagating species adding its own type of monomer to the rate constant for the addition of the other monomer (eqn (5)).

An r₁ value higher than unity implies that preferentially adds an M₁ monomer whereas an r₁ value lower than unity means that preferentially adds an M₂ monomer. Thus, the determination of the reactivity ratios for a monomer couple enables prediction of the reactivity of both monomers during the propagation reaction. Thanks to these reactivity ratios, it is also possible to calculate the evolution of the feed and copolymer composition. Indeed, according to the theory of free radical copolymerization, if f₁ and f₂ are the mole fraction of the monomer M₁ and M₂ in the feed at time t, respectively, and F₁ and F₂ are the mole fractions of the monomer M₁ and M₂ in the copolymer at time t, respectively, then we can write the following equations (eqn (6) and (7)):

The use of the Skeist equation (eqn (8)) allows correlation of the variation of the feed composition with the degree of conversion α.

where

Finally, it is also possible to calculate F_1,c and F_2,c which are the cumulated mole fractions of the monomer M₁ and M₂ in the copolymer at time t, respectively, as shown in eqn (9). Contrary to F₁ and F₂, which represent the mole fractions of monomers M₁ and M₂ in the copolymer at time t, F_1,c and F_2,c values take into account all the copolymer chains created between t = 0 and time t.

By plotting the evolution of F₁, F₂, F_1,c and F_2,c as a function of the conversion α, it is then possible to exactly know the molar composition of the copolymer chains during the polymerization and thus to determine the most probable architecture of the statistical copolymer.

Results and discussion

Copolymers synthesis

We first synthesized statistical copolymers by free radical polymerization using either phosphonated methacrylate or acrylamide monomer (Scheme 1). In both cases, polymerization reaction was achieved under an inert atmosphere using azobisisobutyronitrile (AIBN) as radical initiator (0.5%_mol).

Scheme 1 Synthetic pathway for the synthesis of P(NnPAAm-stat-_hMAPC1), P(NnPAAm-stat-AAmEPA) statistical copolymers, and for the synthesis of P(NnPAAm-block-AAmEP) diblock copolymers.

In order to evaluate the influence of the amount of phosphonated moieties in the copolymer, different NnPAAm monomer/phosphonated monomer ratios (NnPAAm/M_P ratios) were introduced in the feed, ranging from 95/05 to 70/30. All copolymerizations were achieved during a period of 8 hours in order to reach a conversion close to 100%. The copolymer composition was calculated using ¹H NMR spectroscopy, by comparing signals at 4.0–4.3 ppm corresponding to the protons of the methylene α to the phosphorus atom, with those of NnPAAm at 1.2–1.8 ppm attributed to the protons of the methyl group. The copolymers were then recovered after precipitation in cold hexane. As the chemical nature of both phosphonated monomers was different, we did not expect a similar behavior in term of reactivity during the copolymerization reaction, thus leading to the obtaining of different architectures. For instance, in the literature, reactivity ratios for N-methyl acrylamide (MAAm, M₁) and methyl methacrylate (MMA, M₂) were equal to r₁ = 0.05 and r₂ = 1.14, indicating that all propagating chains prefer to react with the MMA, thus leading to a high incorporation of MMA into the copolymer chains.³¹ On the contrary, when copolymerization of N-isopropyl acrylamide (M₁) and N-n-propylacrylamide (M₂) was achieved, reactivity ratios indicated that an alternate copolymer was preferentially obtained (r₁ = 0.6; r₂ = 0.46).³² From these examples, we expected different incorporation of monomers using MAPC1 or DAAmEP, leading to different architectures.

To compare thermal properties of statistical copolymers with well-defined structures, we also prepared block copolymers. P(NnPAAm-block-DAAmEP) was synthesized by RAFT polymerization in two steps. The first step corresponded to the RAFT polymerization of NnPAAm using 2-cyano-2-propyl dodecyl trithiocarbonate (CTA) as chain transfer agent and AIBN as radical initiator. All polymerizations were achieved in DMSO. The target molecular weight was 12 500 g mol⁻¹. P(NnPAAm) was purified by precipitation in distilled water at 45 °C. The second step for the synthesis of P(NnPAAm-block-DAAmEP) consisted of RAFT polymerization of DAAmEP using the P(NnPAAm) previously prepared as macro-chain transfer agent and AIBN as radical initiator. Once again, polymerization was achieved in DMSO. Target molecular weights ranged from 1250 g mol⁻¹ to 10 000 g mol⁻¹ in order to synthesize block copolymers with similar NnPAAm/M_P ratios to those of the statistical copolymers (M_P: phosphoned monomer). The experimental NnPAAm/M_P ratios in the copolymers were determined by ¹H NMR analysis using the same procedure as for other copolymers. P(NnPAAm-block-DAAmEP) 10 000–10 000, 10 000–5000, 10 000–2500, and 10 000–1250 were equivalent to statistical copolymers with NnPAAm/M_P ratios equal to 70/30, 80/20, 90/10, and 95/05, respectively. Experimental molecular weights (M_n,_exp) and dispersities (Đ) were determined by size exclusion chromatography in N,N-dimethylacetamide (DMAc), with 0.1 weight percent of lithium chloride at 50 °C. All corresponding results are summarized in Table 1. In the case of free radical polymerization (entry 1 to 7), 0.5% radical initiator was used. Copolymer chains with molecular weights ranging from 45 000 to 86 000 g mol⁻¹ and high dispersities were obtained in all cases (between 2.6 and 3.8), as expected. On the contrary, RAFT polymerizations (entries 8 to 12) enabled synthesis of block copolymers with controlled molecular weights and dispersity lower than 1.5. All copolymers obtained by RAFT polymerization gave linear evolution of the number-average molecular weight (M_n) vs. conversion (see ESI, Fig. S1†), confirming that termination or transfer reactions were not significant. Additionally, size exclusion chromatography (SEC) measurements (see ESI, Fig. S2†) enabled confirmation that DAAmEP monomers had effectively grown onto the P(NnPAAm) macro-chain transfer agent leading to diblock copolymer architecture. Indeed, the elution times corresponding to P(NnPAAm-block-DAAmEP) block copolymers were lower than for the that of P(NnPAAm) macro-chain transfer agent, indicating an increase in the molecular weight. Results of the 2D diffusion-ordered NMR spectroscopy (2D DOSY NMR) performed on both the P(NnPAAm) macro-chain transfer agent and the P(NnPAAm-block-DAAmEP) copolymers (Fig. 1) were in accordance with SEC. 2D DOSY NMR analyses allow discrimination of macromolecules by their diffusion coefficients.^33,34

Table 1 Characteristics of the statistical and diblock poly(N-n-propylacrylamide)-based copolymers

Entry	Copolymer abbreviation	Mn,th (g mol^−1)	t (min)	Conv^a (%)	Mn,exp^b (g mol^−1)	Đ^b	NnPAAm/MP,exp^a	mmolP/gpoly^c^,^a
a Determined by ^1H NMR analysis.b Determined by size exclusion chromatography in N,N-dimethylacetamide (DMAc), 0.1% LiCl, 50 °C.c Molar amount of phosphonic acid moieties per gram of copolymer.
1	P(NnPAAm-stat-MAPC1) 70/30	—	480	≈100	71 000	2.60	70/30	2.11
2	P(NnPAAm-stat-MAPC1) 80/20	—	480	≈100	45 000	2.82	81/19	1.45
3	P(NnPAAm-stat-MAPC1) 90/10	—	480	≈100	59 100	3.60	92/08	0.68
4	P(NnPAAm-stat-MAPC1) 95/05	—	480	≈100	55 700	3.30	96/04	0.34
5	P(NnPAAm-stat-DAAmEP) 80/20	—	480	≈100	60 700	3.20	79/21	1.48
6	P(NnPAAm-stat-DAAmEP) 90/10	—	480	≈100	86 000	3.60	90/10	0.80
7	P(NnPAAm-stat-DAAmEP) 95/05	—	480	≈100	70 000	3.80	94/06	0.50
8	P(NnPAAm) macro-CTA	10 000	130	78	14 800	1.27	—	—
9	P(NnPAAm-block-DAAmEP) 70/30	24 800	210	81	21 150	1.30	67/33	2.13
10	P(NnPAAm-block-DAAmEP) 80/20	19 800	210	79	19 400	1.30	79/21	1.51
11	P(NnPAAm-block-DAAmEP) 90/10	17 300	210	81	18 000	1.28	88/12	0.94
12	P(NnPAAm-block-DAAmEP) 95/05	16 050	145	79	15 950	1.31	94/06	0.47

---

Fig. 1 2D DOSY NMR spectra of P(NnPAAm) macro-chain transfer agent and P(NnPAAm-block-DAAmEP) diblock copolymers in DMSO-d₆.

The diffusion coefficient values of P(NnPAAm) macro-chain transfer agent and P(NnPAAm-block-DAAmEP) copolymers were measured in DMSO-d₆ and proved to decrease with the size of the P(DAAmEP) second block (from 109.4 m² s⁻¹ for P(NnPAAm) to 57.9 m² s⁻¹ for P(NnPAAm-block-DAAmEP)_{10000–10000}) (Table 2). On the other hand, signals corresponding to P(NnPAAm) and P(DAAmEP) blocks had the same diffusion coefficient values for each copolymer, thereby demonstrating that the P(DAAmEP) block had effectively grown on the P(NnPAAm) macro-chain transfer agent and thus that diblock copolymers were obtained.

Table 2 2D DOSY NMR diffusion coefficient values of P(NnPAAm) macro-chain transfer agent and P(NnPAAm-block-DAAmEP) diblock copolymers

Entry	(Co)polymers	Diffusion coefficients (m^2 s^−1 × 10^−12)	Diffusion coefficients (log(m^2 s^−1 × 10^−12))
1	P(NnPAAm)	109.4	−9.96
2	P(NnPAAm-block-DAAmEP)10000–1250	97.6	−10.01
3	P(NnPAAm-block-DAAmEP)10000–2500	92.9	−10.03
4	P(NnPAAm-block-DAAmEP)10000–5000	82.0	−10.08
5	P(NnPAAm-block-DAAmEP)10000–10000	57.9	−10.24

---

Finally, hydrolysis of dimethyl or diethylphosphoryl functional groups (O=P(OR)₂, R = Me, Et) was achieved using bromotrimethylsilane (BrSiMe₃) and methanol to both (i) obtain the same chemical nature of the phosphorus-based group (i.e. phosphonic acid group) for all copolymers, thus only considering the influence of the chemical nature of the phosphonated co-monomer on the LCST, and (ii) to evaluate the sorption properties of the polyacrylamide-based materials as a function of the architecture. The comparison between the ¹H NMR and ³¹P spectra of the copolymers obtained before and after hydrolysis confirmed that the reaction was efficient. Indeed, in all cases, the signals corresponding to the phosphonated esters (O=P(OMe)₂ in the case of P(NnPAAm-stat-MAPC1) and O=P(OEt)₂ for P(NnPAAm-stat-DAAmEP) and P(NnPAAm-block-DAAmEP)) disappeared, in particular on the ³¹P NMR spectra where no signal at 24 (P(NnPAAm-stat-MAPC1)), 31.2 (P(NnPAAm-stat-MAPC1)) and 29.3 ppm (P(NnPAAm-block-DAAmEP)) was noticeable after hydrolysis.

Three kinds of thermosensitive copolymers bearing phosphonic acid groups were synthesized. Well-defined P(NnPAAm-block-AAmEPA) diblock copolymers were obtained by RAFT polymerization whereas both P(NnPAAm-stat-_hMAPC1) and P(NnPAAm-stat-AAmEPA) were prepared by free radical polymerization. Even if the architecture of the statistical copolymers cannot be controlled, it is possible to evaluate the copolymer chains compositions through the determination of the reactivity ratios of NnPAAm/MAPC1 and NnPAAm/DAAmEP monomer couples. If reactivity ratios are different, the architecture of the statistical copolymer will also differ and, as a consequence, thermal behavior of statistical copolymers will probably be modified.

Determination of the reactivity ratios of statistical copolymers

Even if P(NnPAAm-stat-_hMAPC1) and P(NnPAAm-stat-AAmEPA) have statistical architectures, the composition of both copolymers can be different, depending on the reactivity ratios of the monomers couples.

Reactivity ratios of both (NnPAAm/MAPC1) and (NnPAAm/DAAmEP) monomers couples, listed in Table 3, have been determined by the Jaacks method.³⁵ The reactivity ratios of the (NnPAAm/MAPC1) monomer couple indicates that a propagating chain ending with NnPAAm or MAPC1 monomers preferentially added MAPC1 monomers. In the case of the (NnPAAm/DAAmEP) monomer couple, r₁ is higher than 1 which means that a propagating chain ended by NnPAAm monomer preferentially added NnPAAm monomers. On the contrary, a propagating chain ended by DAAmEP monomer could add DAAmEP monomer and NnPAAm monomer with the same probability as r₂ is almost equal to 1. As these reactivity ratios drive the evolution of the copolymer chains composition during the polymerization, the evolution of the mole fractions of the monomers in the copolymer (F₁ and F₂) as well as the cumulated mole fractions of the monomers in the copolymer (F_1,c and F_2,c) were plotted for both couples. For more clarity, only the copolymers obtained by introducing 80 molar percent of NnPAAm in the feed (f_1,0 = 0.8) were plotted (Fig. 2). For P(NnPAAm-stat-MAPC1) statistical copolymers (Fig. 2a), the first copolymer chains contained 65% and 35% of NnPAAm and MAPC1 monomers, respectively. Such initial composition was explained by (i) the relative amount of NnPAAM monomer in the feed and (ii) the fact that all propagating chains preferentially added MAPC1 monomer. Indeed, MAPC1 monomer was mainly incorporated into the copolymer chains at the beginning of the copolymerization as the propagating chains ended by MAPC1 monomer preferentially added MAPC1 monomers (r₂ = 1.41). Thus, the MAPC1 moieties were likely to be adjacent in the chains at the beginning of the copolymerization process. Then, during the copolymerization, because of the decrease of the relative amount of MAPC1 in the feed, the mole fraction of MAPC1 (F₂) in the copolymer decreased, which led to a significant deviation in the copolymer chain composition. Such deviation finally resulted in the formation of chains that mainly contained NnPAAm at the end of the copolymerization (Scheme 2a). In the case of the P(NnPAAm-stat-DAAmEP) statistical copolymer (Fig. 2b), because both reactivity ratios were close to 1, no significant deviation of the composition was observed during the copolymerization. Copolymers chains created at the beginning of the copolymerization contained 83% and 17% of NnPAAm and DAAmEP monomers, respectively, and DAAmEP moieties were not likely to be adjacent in the copolymer chains. Such a result was expected since (i) the feed mainly contained NnPAAm and (ii) all propagating chains were likely to add NnPAAm monomers. As the conversion increased, the relative amount of NnPAAm in the feed decreased, which explained the slight decrease of the mole fraction of NnPAAm (F₁) in the copolymer. As a result, the cumulated mole fractions of the NnPAAm monomer (F_1,c) and DAAmEP monomer (F_2,c) in the copolymer did not significantly evolve during the copolymerization meaning that the copolymer chains created at the beginning and at the end of the copolymerization had quite similar compositions (Scheme 2b).

Table 3 Reactivity ratios of both (NnPAAm/MAPC1) and (NnPAAm/DAAmEP) monomer couples determined by the Jaacks method

Monomer couple	r1	r2
NnPAAm-MAPC1	0.38	1.41
NnPAAm-DAAmEP	1.26	1.01

---

Fig. 2 Variation of the instantaneous copolymer composition, F₁ and F₂ and the cumulated copolymer composition, F_1,c and F_2,c as function of the reaction conversion for the synthesis of P(NnPAAm-stat-MAPC1) (a) and P(NnPAAm-stat-DAAmEP) (b) copolymers. The dashed lines, corresponding to the instantaneous copolymer composition, and the full lines, corresponding to the cumulated copolymer composition, were calculated from eqn (7) and (9), respectively.

Scheme 2 Representation of the copolymer composition deviation as a function of the reaction conversion for both P(NnPAAm-stat-MAPC1) (a) and P(NnPAAm-stat-DAAmEP) (b) statistical copolymers.

So, the comparison of the evolution of the mole fractions of phosphonated monomer in P(NnPAAm-stat-MAPC1) and P(NnPAAm-stat-DAAmEP) demonstrated that the differences in reactivity ratios led to significant differences in the copolymer architecture, having an impact on the thermosensitive behavior of such copolymers in aqueous solution. By playing on the chemical nature of the phosphonated monomer (methacrylate or acrylamide), it will probably be possible to vary the CP value, using copolymers prepared by a very easy polymerization process compared to controlled radical polymerization, for instance.

Influence of the copolymer architecture on the thermo-responsive properties

Determination of cloud points was achieved for both statistical and diblock copolymers (Table 4). The transmittance measurements were carried out on 5 g L⁻¹ copolymer solutions in water. First, results shown in both Fig. 3 and Table 4 demonstrate that for a same copolymer series the cloud points logically increased with the proportion of phosphonic acid groups as the latter are hydrophilic. For instance, in the case of P(NnPAAm-stat-AAmEPA) copolymer, the CPs increased from 24.6 to 28.3 °C when the molar concentration of phosphonated monomer in the copolymer increased from 5%_mol (95/05) to 10%_mol (90/10).

Table 4 Cloud points values of P(NnPAAm-stat-AAmEPA), P(NnPAAm-stat-_hMAPC1) and P(NnPAAm-block-AAmEPA) copolymers containing different relative amount of phosphonic acid moieties (copolymer concentration: 5 g L⁻¹)

Copolymer	Copolymer composition (NnPAAm/phosphonated monomer)
	95/05	90/10	80/20	70/30
P(NnPAAm-stat-AAmEPA)	24.6 ± 1 °C	28.3 ± 1.5 °C	—	—
P(NnPAAm-stat-hMAPC1)	22.8 ± 1 °C	23.4 ± 1.5 °C	25.6 ± 2 °C	—
P(NnPAAm-block-AAmEPA)	22.1 ± 1.5 °C	22.9 ± 1.5 °C	23.5 ± 2 °C	24.8 ± 3 °C

---

Fig. 3 Plot of transmittance as a function of the temperature for aqueous solutions of P(NnPAAm-stat-AAmEPA) (a), P(NnPAAm-stat-_hMAPC1) (b) and P(NnPAAm-block-AAmEPA) (c) copolymers with different ratios of phosphonic acid moieties (from 5%_mol (95/05) to 30%_mol (70/30)). Copolymer concentration: 5 g L⁻¹; ramp temperature: 0.2 °C min⁻¹.

P(NnPAAm-stat-AAmEPA), containing a higher amount of phosphonic acid moieties, did not show thermosensitive behavior. For P(NnPAAm-stat-_hMAPC1), increasing the amount of phosphonic acid also led to an increase of the CPs. The latter increased from 22.8 to 25.6 °C going from P(NnPAAm-stat-_hMAPC1) 95/05 to 80/20 and the copolymer containing 30%_mol of phosphonated monomer (70/30) did not show well-defined CPs values as the copolymer almost lost its thermosensitivity. Finally, the increase of the cloud points values was less pronounced in the case of well-defined P(NnPAAm-block-AAmEPA) diblock copolymer. Indeed, the LCST increased from 22.1 to 24.8 °C from P(NnPAAm-block-AAmEPA) 95/05 to 70/30. All P(NnPAAm-block-AAmEPA) samples had similar thermosensitive behavior even if P(NnPAAm-block-AAmEPA) 70/30 was characterized by a wide transition of solubility. Transitions were relatively sharp whatever the copolymer when the phosphonic acid content remained less than 20%. Sharpest transitions were logically obtained from polymers prepared by RAFT due to the uniformity of the chain length.

Aside from the phosphonic acid group content, it is obvious that the copolymer architecture has also an influence on the cloud point values. It is important to mention that even if statistical and diblock copolymers had different molecular weights, the influence of the polymer molar mass onto the cloud point appeared negligible.³⁶ Results reported in Table 4 show that for an equivalent relative amount of phosphonic acid moieties in the copolymer, the CP increased with the dispersion of these hydrophilic groups along the copolymer chains. For P(NnPAAm-stat-AAmEPA), the presence of phosphonic acid groups on the whole copolymer chain means that the acrylamide moieties are surrounded by phosphonic acid groups. Thus, the hydrogen bonds broke and the creation of polymer–polymer interactions was difficult due to the presence of these highly hydrophilic groups nearby the acrylamide monomers. This explains the relatively high LCST values obtained for P(NnPAAm-stat-AAmEPA) copolymers. On the contrary, in the case of P(NnPAAm-block-AAmEPA), the phosphonic acid moieties were all located side by side along the macromolecular chains due to the diblock architecture and thus did not influence significantly the interaction between the acrylamide moieties and water. As a result, the CP of the P(NnPAAm-block-AAmEPA) diblock copolymer did not evolve a lot with the increasing amount of phosphonic acid moieties. Finally, P(NnPAAm-stat-_hMAPC1) had intermediate behavior, which was logical since, as shown by the determination of the reactivity ratios, the phosphonic acid moieties dispersion in P(NnPAAm-stat-_hMAPC1) was located between those of P(NnPAAm-stat-AAmEPA) and P(NnPAAm-block-AAmEPA). As a conclusion, the thermosensitivity of the copolymers was affected not only by the amount of phosphonic acid hydrophilic groups but also by the copolymer architecture. The more the hydrophilic moieties were grouped together on the copolymer chains, the less they affected the interactions between the acrylamide moieties and water and thus the lower the cloud point was. So, it is possible to adapt the cloud points of statistical copolymers knowing the reactivity ratios of the monomers used for the synthesis. This is of great interest as free radical polymerization is a versatile polymerization process.

Determination of the sorption properties as a function of the copolymers architecture

Thermosensitive copolymers bearing acid functions proved to be useful for water treatment applications. In particular, phosphonic acid groups showed good sorption properties at temperatures lower than the cloud point and good polymer–water separation above the CP since the polymeric sorbent becomes non-soluble.^8,9 Until now, to our knowledge, no special attention was given in the literature to the influence of the copolymeric sorbent architecture on the thermosensitive properties of statistical copolymers. But this point is of interest as for an industrial development of such materials, it would be obviously better to achieve synthesis by free radical polymerization. As previously demonstrated, the architecture of thermosensitive copolymers containing phosphonic acid groups can significantly affect the copolymer behavior in water. So, it was interesting to also focus on the influence of the copolymer architecture on the sorption properties. The three different copolymers were tested. The sorption properties for nickel (Ni) cations were evaluated at various Ni/P ratios. The latter represents the molar amount of Ni introduced in solution compared to the molar amount of phosphonic acid sorption moieties involved in the sorption. All experiments were carried out at 20 °C for 24 hours using the same method as described in a previous paper.⁹

Fig. 4 reports the variation of the molar amount of Ni sorbed per mole of phosphonic sorption moieties (mmol_Ni,sorbed/mmol_P) for Ni/P ratios ranging from 0 to 6 for the three aforementioned copolymers. All copolymers exhibited similar sorption properties: the number of Ni ions sorbed per phosphonic acid sorption sites increased from 0 to 1 when increasing Ni/P. Different coordination modes could explain the interactions between phosphonic acid functional groups and the Ni divalent cations. At low Ni/P ratios, i.e. when the amount of phosphonic acid sorption moieties was in excess compared to the amount of metallic pollution, the sorption capacity was mainly explained by interactions involving an average of two sorption sites per cation.

Fig. 4 Sorption capacities (mmol_Ni,sorbed/mmol_P) of P(NnPAAm-stat-AAmEPA), P(NnPAAm-stat-_hMAPC1) and P(NnPAAm-block-AAmEPA) copolymers for Ni²⁺ cations for different Ni/P ratios.

Under such conditions, a significant amount of phosphonic acid moieties were not involved in the sorption process, whereas at high Ni/P ratios, all the phosphonic acid groups were involved in the sorption process, meaning that the main coordination modes were different: the phosphonic acid–Ni interactions involved an average of one sorption site per cation.⁸ The similarity of the sorption properties in the case of the three copolymers highlighted that the coordination modes were not affected by the copolymer architecture. Hence, the proximity of the sorption sites into the copolymer chains did not have any significant impact on the sorption either at low or high Ni/P ratio. This relationship between the copolymer architecture and the sorption properties is not very surprising since we demonstrated in a previous paper that the interaction between phosphonic acid and divalent cations mainly resulted from ion exchange interactions rather than from complexation.³⁷ Indeed, if the sorption phenomena were driven by electrostatic interactions, the main parameter affecting the sorption capacity would be the global charge of the copolymer rather than the localization of these charges on the copolymer chains. Then, the fact that these three copolymers characterized by three different architectures had the same sorption capacities was in agreement with a sorption process based on electrostatic interactions. By employing a specific material, a particular cloud point will be obtained, depending on the copolymer architecture. As sorption was the same in all cases, the choice of the copolymer will be driven by the obtaining of the lowest increase of temperature of the industrial effluent to go from soluble to non soluble state. So, by varying the phosphonated-based monomer, it will be possible to adapt the cloud point value of the statistical copolymers, using a polymerization process that is easy to carry out.

Conclusions

New thermosensitive copolymers containing phosphonic acid moieties and having different architectures were synthesized. Thanks to the free radical polymerization technique, it was possible to synthesize both P(NnPAAm-stat-AAmEPA) and P(NnPAAm-stat-_hMAPC1) statistical copolymers. The copolymer architecture was not controlled but the determination of the reactivity ratios of (NnPAAm/MAPC1) and (NnPAAm/DAAmEP) monomer couples enabled to highlight differences between these two copolymers. For P(NnPAAm-stat-AAmEPA) copolymers, the reactivity ratios of both monomers were very similar, which meant that all polymer chains had similar composition, whatever the polymerization conversion. For P(NnPAAm-stat-_hMAPC1), the reactivity of both monomers were different, which led to a significant composition deviation between the copolymer chains created at the beginning and at the end of the polymerization. As a consequence, the phosphonic acid groups were more dispersed along the polymer chains for P(NnPAAm-stat-AAmEPA) than for P(NnPAAm-stat-_hMAPC1). To complete this study, P(NnPAAm-block-AAmEPA) diblock copolymers were also prepared by RAFT polymerization. Whatever the polymerization technique used, it was possible to control the relative molar amount of phosphonic acid introduced into the copolymer chains, which permitted evaluation of the influence of the phosphonic acid content on the copolymer properties, and more importantly the influence of the localization of these moieties along the copolymer chains. The main conclusions were that the cloud points of the copolymer increased (i) with the number of the phosphonic acid groups for a given architecture, as expected, and (ii) with the dispersion of the phosphonic acid on the copolymer chains for a given content of phosphonic acid moieties. Additionally, the sorption properties of all three copolymers were evaluated. No significant differences were observed, meaning that the copolymer architecture did not have any influence on the sorption.

To conclude, it is possible to adapt cloud points of statistical poly(N-n-propyl acrylamide)-based copolymers judiciously by choosing the chemical nature of the co-monomer (methacrylate or acrylamide) used for the synthesis. Indeed, by using either DAAmEP or MAPC1 to prepare statistical copolymers by free radical polymerization, we showed that different reactivity ratios and thus copolymer architectures were obtained, leading to a significant variation of the cloud points values. Additionally, P(NnPAAm-stat-_hMAPC1) easily prepared by free radical polymerization had comparable thermosensitive properties to P(NnPAAm-block-AAmEPA) diblock copolymers synthesized by RAFT. As a consequence, as sorption properties remained unchanged whatever the architecture, these results demonstrate that it is possible to adapt the cloud point to the temperature of an industrial effluent to minimize the thermal exchange by choosing the appropriate phosphonated comonomer while keeping a versatile polymerization process.

Acknowledgements

The authors thank the “Agence Nationale de la Recherche” (ANR) for funding this work via the ANR COPOTERM under the 2009 ECOTECH program (ANR-09-ECOT-005-02).

Notes and references

1. R. X. Liu, M. Fraylich and B. R. Saunders, Colloid Polym. Sci., 2009, 287, 627–643.
2. H. G. Schild, Prog. Polym. Sci., 1992, 17, 163–249.
3. C. D. H. Alarcon, S. Pennadam and C. Alexander, Chem. Soc. Rev., 2005, 34, 276–285.
4. S. Monge, S. Antoniacomi, V. Lapinte, V. Darcos and J. J. Robin, Polym. Chem., 2012, 3, 2502–2507.
5. B. R. Saunders, N. Laajam, E. Daly, S. Teow, X. H. Hu and R. Stepto, Adv. Colloid Interface Sci., 2009, 147–148, 251–262.
6. M. D. Kurkuri, M. R. Nussio, A. Deslandes and N. H. Voelcker, Langmuir, 2008, 24, 4238–4244.
7. H. Vihola, A. K. Marttila, J. S. Pakkanen, M. Andersson, A. Laukkanen, A. M. Kaukonen, H. Tenhu and J. Hirvonen, Int. J. Pharm., 2007, 343, 238–246.
8. A. Graillot, D. Bouyer, S. Monge, J.-J. Robin and C. Faur, J. Hazard. Mater., 2013, 244–245, 507–515.
9. A. Graillot, S. Djenadi, C. Faur, D. Bouyer, S. Monge and J. J. Robin, Water Sci. Technol., 2013, 67, 1181–1187.
10. T. Nonaka, Y. Hanada, T. Watanabe, T. Ogata and S. Kurihara, J. Appl. Polym. Sci., 2004, 92, 116–125.
11. H. Tokuyama, K. Yanagawa and S. Sakohara, Sep. Purif. Technol., 2006, 50, 8–14.
12. I. Dimitrov, B. Trzebicka, A. H. E. Muller, A. Dworak and C. B. Tsvetanov, Prog. Polym. Sci., 2007, 32, 1275–1343.
13. L. Hanykova, J. Spevacek and M. Ilavsky, Polymer, 2001, 42, 8607–8612.
14. R. R. Bhattacharjee, M. Chakraborty and T. K. Mandal, J. Phys. Chem. B, 2006, 110, 6768–6775.
15. O. F. Solomon, M. Corciove, I. Ciuta and C. Boghina, J. Appl. Polym. Sci., 1968, 12, 1835–1842.
16. J. Ramos, A. Imaz and J. Forcada, Polym. Chem., 2012, 3, 852–856.
17. J. Spevacek, J. Dybal, L. Starovoytova, A. Zhigunov and Z. Sedlakova, Soft Matter, 2012, 8, 6110–6119.
18. M. Pakravan, M. C. Heuzey and A. Ajji, Macromolecules, 2012, 45, 7621–7633.
19. E. E. Dormidontova, Macromolecules, 2004, 37, 7747–7761.
20. V. Aseyev, H. Tenhu and F. M. Winnik, Adv. Polym. Sci., 2011, 242, 29–89.
21. K. Van Durme, G. Van Assche, V. Aseyev, J. Raula, H. Tenhu and B. Van Mele, Macromolecules, 2007, 40, 3765–3772.
22. A. Miasnikova and A. Laschewsky, J. Polym. Sci., Part A: Polym. Chem., 2012, 50, 3313–3323.
23. M. Luzon, C. Boyer, C. Peinado, T. Corrales, M. Whittaker, L. Tao and T. P. Davis, J. Polym. Sci., Part A: Polym. Chem., 2010, 48, 2783–2792.
24. M. A. Ward and T. K. Georgiou, J. Polym. Sci., Part A: Polym. Chem., 2013, 51, 2850–2859.
25. C. Boyer, M. H. Stenzel and T. P. Davis, J. Polym. Sci., Part A: Polym. Chem., 2011, 49, 551–595.
26. A. Graillot, S. Monge, C. Faur, D. Bouyer and J. J. Robin, Polym. Chem., 2013, 4, 795–803.
27. J. Kohler, H. Keul and M. Moller, Chem. Commun., 2011, 47, 8148–8150.
28. V. Zoulalian, S. Zurcher, S. Tosatti, M. Textor, S. Monge and J. J. Robin, Langmuir, 2010, 26, 74–82.
29. R. Tayouo, G. David, B. Ameduri, J. Roziere and S. Roualdes, Macromolecules, 2010, 43, 5269–5276.
30. G. G. Odian, Principles of Polymerization, Wiley, New York, 3rd edn, 1991.
31. J. Brandrup, E. H. Immergut and E. A. Grulke, Polymer Handbook, John Wiley & Sons, Hoboken, New Jersey, 4th edn, 1999.
32. M. K. Kokufuta, S. Sato and E. Kokufuta, Colloid Polym. Sci., 2012, 290, 1671–1681.
33. Y. Bakkour, V. Darcos, S. M. Li and J. Coudane, Polym. Chem., 2012, 3, 2006–2010.
34. B. Guillerm, V. Darcos, V. Lapinte, S. Monge, J. Coudane and J. J. Robin, Chem. Commun., 2012, 48, 2879–2881.
35. V. Jaacks, Macromol. Chem., 1972, 161, 161–172.
36. A. Housni and R. Narain, Eur. Polym. J., 2007, 43, 4344–4354.
37. A. Graillot, D. Bouyer, S. Monge, J. J. Robin, P. Loison and C. Faur, J. Hazard. Mater., 2013, 2013, 425–433.

---

Footnote

† Electronic supplementary information (ESI) available: Information about the synthesis of the copolymers and method used for the determination of the sorption properties. See DOI: 10.1039/c4ra00140k

---