Polymer Chemistry

Polymer Chemistry PAPER

Cite this: Polym. Chem., 2018, 9, 646

Electrochemically mediated ATRP in ionic liquids: controlled polymerization of methyl acrylate in [BMIm][OTf ]† Francesco De Bon,a Marco Fantin,

b

Abdirisak A. Isse

*a and Armando Gennaro

*a

Electrochemically mediated atom transfer radical polymerization (eATRP) of methyl acrylate (MA) was carried out for the first time in an ionic liquid, 1-butyl-3-methylimidazolium trifluoromethanesulfonate ([BMIm][OTf ]). Ionic liquids are particularly suitable as solvents for eATRP because (i) a supporting electrolyte is not required, (ii) purification of the polymer is simple, and (iii) the catalyst can be recycled maintaining most of its original activity. Polymerization was relatively fast and well-controlled, affording polymers with a dispersity of ∼1.1 at >90% conversion in 2 h. Different conditions were explored, including applied potential, applied current, temperature, the monomer/solvent ratio, targeted degree of polymerization (from 276 to 828), and catalyst loading (down to 45 ppm). The livingness of MA polymerization was Received 27th December 2017, Accepted 8th January 2018

confirmed by realizing an electrochemical switch (ON/OFF toggling) and by the chain extension of poly (methyl acrylate)-Br with acrylonitrile. Chain extension was carried out for the first time by a “catalytic

DOI: 10.1039/c7py02134h

halogen exchange”, i.e. 180 ppm of a Cu complex was used to convert the chain end from C–Br to C–Cl.

rsc.li/polymers

This afforded a well-controlled poly(methyl acrylate)-b-poly(acrylonitrile)-Cl copolymer.

Introduction Reversible deactivation radical polymerizations (RDRPs) are powerful techniques for macromolecular engineering, allowing the preparation of tailor-made polymers with well-defined composition, architecture and topology.1 Among RDRP techniques, atom transfer radical polymerization (ATRP) is the most versatile. It has been successfully applied to a vast range of monomers: (meth)acrylates,2–5 methacrylic acid,6 acrylamides,7–9 4-vinylpyridine,10,11 styrene,4,12,13 and vinyl chloride.14,15 In an ATRP process, a dormant polymer is activated by a transition metal complex (often a copper–amine complex) to generate radicals via atom transfer (Scheme 1). Simultaneously, the transition metal is oxidized to its higher oxidation state, which can then deactivate the growing radicals. This reversible activation/deactivation process establishes an equilibrium that is predominantly shifted to the dormant chains; therefore, the radical concentration is very low. The fast activation/ deactivation equilibrium allows preparation of polymers with

narrow molecular weight distribution and pre-determined molecular weights. The ratio of monomer to polymerization initiator (alkyl halide, RX) determines the number of growing chains and the degree of polymerization (DP). One of the major disadvantages of the original ATRP procedure was the requirement of high loading of a metal catalyst, which must be removed after polymerization using time-consuming and expensive methods.16 New ATRP techniques have been developed to reduce catalyst loading, namely activators regenerated by electron transfer (ARGET) ATRP,16 supplemental activator and reducing agent (SARA) ATRP17–19 also improperly named SET-LRP,20,21 initiators for continuous activator regeneration (ICAR) ATRP,22 photoinduced ATRP,23–25 electrochemically mediated ATRP (eATRP),26,27 and mechanically mediated ATRP.28

a

Department of Chemical Sciences, University of Padova, Via Marzolo 1, 35131 Padova, Italy. E-mail: [email protected], [email protected] b Department of Chemistry, Carnegie Mellon University, 4400 Fifth Avenue, Pittsburgh, PA, 15213, USA † Electronic supplementary information (ESI) available: Experimental details, comparison of eATRP in ILs and acetonitrile, galvanostatic eATRP and characterization of purified and recycled ILs. See DOI: 10.1039/c7py02134h

646 | Polym. Chem., 2018, 9, 646–655

Scheme 1 Mechanism of eATRP with electrochemical regeneration of the active catalyst.

This journal is © The Royal Society of Chemistry 2018

Polymer Chemistry

eATRP uses current, flowing through a working electrode, to (re)generate the active [CuIL]+ catalyst (Scheme 1, where L = amine ligand). eATRP is an attractive technique from an environmental point of view because of the absence of byproducts in the polymerization mixture, in comparison with other low-ppm techniques such as ARGET, ICAR and SARA ATRP. In eATRP, undesired products are generated in the anodic compartment, which is separated from the cathodic compartment where polymerization occurs.26 eATRP, however, has some limitations. A supporting electrolyte is typically required to achieve suitable conductivity in aqueous or organic solvents. Moreover, like other ATRP techniques, the employed organic solvents and copper catalysts are generally toxic29 and expensive to recycle or dispose of. Ionic liquids (ILs), typically composed of an organic cation and a weakly coordinating anion, are an attractive alternative to common organic solvents. They are non-volatile, usually non-flammable, electrically conductive, thermally stable up to >300 °C, and easily recyclable. In addition, they are generally good solvents for inorganic, organic, and polymeric materials.30,31 ATRP in ILs has been explored to facilitate polymer separation and catalyst recycling. Carmichael et al.32 reported the first ‘normal ATRP’ of methyl methacrylate (MMA) in [BMIm][PF6], catalysed by CuIBr/N-propyl-2-pyridylmethyl amine with ethyl 2-bromoisobutyrate as an initiator. The polymerization rate was comparable to that in common organic solvents. Taking advantage of the insolubility of copper complexes in toluene and the immiscibility of the latter with [BMIm][PF6], the polymer was successfully separated by extraction using toluene. Afterwards, several ATRPs were conducted in a series of ILs with the same PF6− anion, all able to solubilize monomers such as MMA and acrylonitrile (AN).33 [BMIm][PF6] was also used for the sequential ATRP of butyl acrylate and methyl acrylate providing block copolymers with very low dispersity (Đ < 1.13).31 This ionic liquid was also used to prepare the copolymer poly(ethylene oxide)-b-polystyrene.34 ARGET ATRP of MMA and AN was performed in several ILs.33,35 ILs have never been used as solvents for eATRP, despite being well suited for applications in electrochemical techniques. Indeed, ILs have a broad electrochemical window and sufficient conductivity, so that no external supporting electrolyte is necessary, leading to a cleaner polymerization system. An eATRP system in an IL is composed of only four elements: an IL solvent, a monomer, an initiator, and a small amount of catalyst with continuous regeneration of the activator [CuIL]+ by electrochemical reduction. The number of possible ILs is high; more than 1000 ILs are described in the literature and 300 are commercially available.36 Selection of an appropriate IL should avoid the negative impact of impurities on ATRP: (i) nitrogen-containing byproducts from the solvent synthesis may bind to the copper complex reducing its activity or selectivity; (ii) halide ions decrease the rate of ATRP37 and reduce the available electrochemical potential window; (iii) water narrows the electro-


Paper

chemical window and may also hydrolyse certain C–X functionalities (e.g. benzyl bromide); and (iv) acidic impurities can protonate alkyl amine ligands such as N,N,N′,N″,N″-pentamethyldiethylenetriamine (PMDETA) and tris[2-(dimethylamino) ethyl]amine (Me6TREN), leading to decomposition of the catalyst.38 In addition, the viscosity of the IL should be low enough to allow effective mass transport to and from the electrode surface. 1-Butyl-3-methylimidazolium trifluoromethanesulfonate, [BMIm][OTf ], was used because of its adequate viscosity (2-fold lower than that of [BMIm][BF4] and 5-fold lower than that of [BMIm][PF6])30 and a wide electrochemical window (from −2.4 V to +1.4 V vs. Fc+|Fc, see Fig. S1†). Moreover, BMIm-based ILs have already been successfully used in ATRP.35 We have recently reported that copper catalysts with traditional ligands (PMDETA, Me6TREN, and tris(2-pyridylmethyl)amine (TPMA)) and halide ions are stable in [BMIm][OTf ].39 The reactivity of the catalysts depends on the ligand, in the following order PMDETA < TPMA < Me6TREN for [XCuIIL]+. In general, Eoˉ of [XCuIIL]+ is more negative than that of [CuIIL]2+, indicating that X− stabilizes more Cu(II) than Cu(I), which is a crucial requisite of a good ATRP catalyst. Moreover, activation of ATRP initiators in [BMIm][OTf ] by [CuITPMA]+ showed rate constants similar to those measured in traditional molecular solvents such as acetonitrile. Also, the effects of the RX molecular structure and type of halogen on the activation rate were the same as in organic solvents.39,40 This further suggested that [BMIm][OTf ] is a good candidate for well-controlled eATRP. We describe herein a polymerization system with the following characteristics: (i) use of an IL as a green solvent, (ii) low catalyst loading ( ppm levels of CuBr2/TPMA), (iii) electrochemical (re)generation of the active catalyst, thus avoiding chemical reducing agents and associated byproducts, (iv) simple extraction of the produced polymer, and (v) a facile catalyst and IL recycling. Moreover, preservation of the chain-end functionality was investigated by chain extension of a poly(methyl acrylate)-Br macroinitiator with acrylonitrile. In this context, the –Br chain end was converted to –Cl using only a catalytic amount of the CuII complex in a ‘catalytic halogen exchange’. This is a significant improvement of the traditional halogen exchange procedure, which until now required an equimolar amount of the air-sensitive CuI complex with respect to the –Br chain end.41,42

Experimental Materials and methods 1-Butyl-3-methylimidazolium trifluoromethanesulfonate ([BMIm][OTf ], Iolitec, 99%) was first neutralized to pH = 7 using aqueous 0.2 M KOH (50 mL of [BMIm][OTf ] were dissolved in 50 mL of double-distilled water and then titrated to pH = 7). Water was then removed using a rotary evaporator. The IL was vigorously stirred for 180 min with 50 mL of diethyl

Polym. Chem., 2018, 9, 646–655 | 647

Paper

ether, after which the mixture was allowed to settle and then separated. After repeating this procedure once more, a third aliquot of 100 mL of diethyl ether was added to the IL, followed by intense stirring for 12 h and separation. Residual ether in the IL phase was removed using a rotary evaporator and then the ionic liquid was dried under high vacuum (0.03 mbar) at 110 °C for at least 24 h. The yield was ca. 95%. CuBr2 (Sigma-Aldrich, 99.999% trace metal basis), tris(pyridylmethyl)amine (TPMA, Sigma-Aldrich, 97%), ethyl α-bromoisobutyrate (EBiB, Aldrich, 98%), H2SO4 (Fluka, 95% for analysis), toluene (VWR, 99%), and deuterated chloroform (99.8 atom% D, Sigma-Aldrich) were used as received. Copper complexes were prepared in situ by mixing appropriate aliquots of stock solutions of a copper salt and TPMA in the purified ionic liquid. Methyl acrylate (MA) (Sigma-Aldrich, >99%) and acrylonitrile (AN, Sigma Aldrich >99%) were purified by passing through a column of basic alumina to remove polymerization inhibitors; purified monomers were stored at 4 °C in the dark. Tetraethylammonium chloride (Sigma Aldrich, >98%), used as a source of chloride ions, was recrystallized from acetone. Acetonitrile (CH3CN, Sigma-Aldrich 99%) was purified by distillation over CaH2 and stored under an argon atmosphere. Deionized water was double-distilled in the presence of KMnO4 to remove possible organic contaminations. Instrumentation Electrochemical measurements and polymerizations were performed with an Autolab PGSTAT 30 potentiostat/galvanostat in a thermostated five-neck cell under a N2 atmosphere; the electrodes were a Pt mesh (estimated geometrical area: ∼6 cm2) or glassy carbon (GC) used as a working electrode (WE), a Pt counter electrode, and an Ag|AgNO3 (0.1 M in [BMIm][OTf ]) reference electrode. Ferrocene was used as an internal standard to calibrate the reference electrode, which allowed conversion of all potentials to the ferrocenium/ferrocene scale. Prior to each experiment, the GC electrode – a 3 mm disc used for voltammetric investigations – was polished with 0.25 µm diamond paste and then sonicated in ethanol for 5 min. Polymer molecular weights and dispersity (Đ) were determined by gel permeation chromatography (GPC). The GPC system was an Agilent 1260 Infinity chromatograph equipped with a refractive index (RI) detector and two PLgel Mixed-D columns (300 mm and 5 μm) connected in series. The column system was calibrated with PMMA standards (molecular weights = 694 to 1 944 000). DMF was used as an eluent, with the column compartment thermostated at 70 °C and RI detector at 50 °C. Samples were filtered with a 200 nm PTFE filter and neutral alumina prior to injection. Conversion was calculated from the 1H NMR spectra of the reaction mixture, recorded on a Bruker Ultrashield 300 MHz spectrometer in CDCl3; the signals of aromatic hydrogens of [BMIm][OTf ] were used as an internal standard. Typical eATRP procedure A five-neck, water-jacketed electrochemical cell continuously flushed with N2 was loaded with 500 µL of 10−2 M CuBr2 stock

648 | Polym. Chem., 2018, 9, 646–655

Polymer Chemistry

solution in [BMIm][OTf ] (0.005 mmol CuBr2), 500 µL of 10−2 M TPMA stock solution in [BMIm][OTf ] (0.005 mmol TPMA), and 1.5 mL (6.72 mmol) of the IL. After 20 min of degassing, 2.25 mL of methyl acrylate (MA) were added and the gas stream was reduced to avoid monomer evaporation. After 10 min of further degassing, the cell was slightly pressurized and sealed. Then, the temperature of the cell jacket was adjusted and, after reaching the selected temperature, 250 µL of a degassed solution of initiator EBiB (0.05 mmol) in MA were injected into the cell. Lastly, the desired potential or current was applied to trigger polymerization. During the reaction, samples for GPC and NMR analysis were periodically withdrawn by a syringe filled with N2.

Results and discussion Electrochemical characterization of the catalyst Cyclic voltammetry (CV) of the purified IL showed a wide potential window without any faradaic current attributable to the oxidation or reduction of impurities (Fig. S1†). Prior to each polymerization, stability and electrochemical properties of the system were evaluated by cyclic voltammetry (Fig. 1). [BrCuIITPMA]+, formed in situ, exhibited a quasireversible peak couple, with E1/2 = (Epc + Epa)/2 = −0.62 V vs. Fc+|Fc, where Epc and Epa are the cathodic and anodic peak potentials, respectively. ˉ o ½BrCuII Lþ þ e Ð ½BrCuI L E½BrCu II þ L =½BrCuI L

ð1Þ

A similar voltammetric pattern was observed when MA was added to prepare the typical IL/MA mixture (1 : 1, v/v) used in eATRP (Fig. 1, trace b). The effect of the monomer is a positive shift of E1/2 to −0.57 V vs. Fc+|Fc and an increase in current intensity. The positive shift of E1/2 may be attributed to the lower polarity of the IL/MA mixture compared to pure IL. Similar positive shifts of E1/2 of copper catalysts have been previously reported in molecular solvents after monomer addition.43,44 Instead, the current enhancement arises from the lower viscosity of the mixture compared to the IL, despite dilution due to the added monomer.

Fig. 1 Cyclic voltammetry of 2 × 10−3 M [BrCuIITPMA]+ in [BMIm][OTf ] (a) or 10−3 M [Br–CuIITPMA]+ in MA/[BMIm][OTf] (1 : 1, v/v) in the absence (b) and in the presence of 10−2 M EBiB (c), recorded at 50 °C on a GC disk electrode at 0.1 V s−1.


Polymer Chemistry

Paper

Reversibility observed in CV indicated that both CuI and Cu were stable in both IL and IL + MA (1 : 1, v/v). Once the initiator was added, the voltammetric pattern changed: the cathodic peak increased while the anodic one almost disappeared, indicating the occurrence of electrocatalysis. On the electrode surface, [BrCuIIL]+ was reduced to [BrCuIL]+, which partially dissociated generating the active form of the catalyst, [CuIL]+.45 The latter reacted with the polymerization initiator, forming an alkyl radical and the oxidized catalyst species [BrCuIIL]+, which diffused back to the electrode to be reduced again to [BrCuIL]. Consequently, the cathodic peak became catalytic while the anodic one decreased in intensity because of the disappearance of CuI via the reaction with RX. The standard reduction potential of the complex was assumed to be identical to the half-wave potential (E1/2) of the redox couple, assuming similar diffusion coefficients for the oxidized (DCuII) and reduced (DCuI) species and the activity coefficient ratio (γCuII/γCuI) close to unity (eqn (2)): II

E1=2 ¼ Eˉo þ

RT γ CuII ln nF γ CuI

RT D I 1=2 ln Cu þ nF DCuII

ð2Þ

Fig. 2 Effect of Eapp on eATRP of 50% (v/v) MA in [BMIm][OTf ]. (a) Firstorder kinetic plots and (b) evolution of Mn and Đ with conversion. Conditions: CMA : CCuBr2 : CTPMA : CRX = 552 : 1 : 0.1 : 0.1. CCuBr2 = 10−3 M. The dotted line represents the theoretical Mn.

where R is the gas constant and F is the Faraday constant. The values of applied potential (Eapp) used during eATRP were chosen on the basis of E1/2 measured for [BrCuIITPMA]+ reduction in MA/[BMIm][OTf ] (1 : 1, v/v).

Electrochemically mediated atom transfer radical polymerization Effect of applied potential. A series of eATRPs of 50% (v/v) MA in [BMIm][OTf ] were performed at different Eapp values, from E1/2 to E1/2 − 0.12 V (Table 1 and Fig. 2). A catalyst concentration of 10−3 M was chosen for these experiments, which is typical for electrochemical investigations and corresponds to 180 ppm of the catalyst (molar ratio to MA). The ratio between CuI and CuII concentrations at the electrode surface is roughly related to Eapp, according to the following equation: Eapp ¼ E1=2 þ

RT CCuII ln F CCuI

ð3Þ

Therefore, changing Eapp is a straightforward way to modulate the rate (Rp) of polymerization, which strongly depends on the CuI to CuII ratio:

Table 1

RP ¼

dCM CRX CCuI CM ¼ kpapp CM ¼ kp KATRP dt CCuII

ð4Þ

where kp is the propagation rate constant, KATRP is the ATRP equilibrium constant, CM is the monomer concentration and app kp ¼ kp CR• is the apparent propagation rate constant. As expected, lowering the Eapp from E1/2 to E1/2 − 0.12 V induced faster CuI (re)generation, thus enhancing the polymerization app rate; kp increased from 0.17 h−1 to 0.79 h−1. On the other hand, lowering the Eapp decreased the concentration of the deactivating CuII species and therefore hampered polymerization control. For example, at Eapp = E1/2 − 0.12 V, where the target ratio between the deactivating and activating species was 1/100 (see eqn (3)), a fast but only moderately controlled polymerization (Đ = 1.32) was observed (Table 1, entry 5). Despite the slight deterioration of control with decreasing Eapp, the measured Mn was always in excellent agreement with theoretical values (Mn,th) (Fig. 2b). Fig. 3 shows the effect of Eapp on conversion and Đ for eATRP of MA in [BMIm][OTf ]. The best conditions were found at Eapp = E1/2 − 0.06 V: 90% conversion in 5 h, Mn in excellent

eATRP of 50% (v/v) MA in [BMIm][OTf ] at different Eapp values, T = 25 °Ca app c

Entry

Eapp − E1/2 (V)

CCuII/CCuI b

t (h)

Conversion (%)

kp

1 2 3 4 5

0 −0.03 −0.06 −0.09 −0.12

1 0.3 0.1 0.03 0.01

5 5 5 4 3.5

59 70 90 94 96

0.17 0.26 0.48 0.61 0.79

(h−1)

Mn,th

Mn,app

Đ

28 000 33 300 42 800 44 700 46 600

27 300 32 300 42 300 42 900 44 100

1.08 1.10 1.10 1.24 1.32

Conditions: CMA : CCuBr2 : CTPMA : CRX = 552 : 0.1 : 0.1 : 1, CCuBr2 = 10−3 M; Vtot = 5 mL. b Calculated from eqn (3). c The slope of the ln([M]0/[M]) vs. t plot.

a


Polym. Chem., 2018, 9, 646–655 | 649

Paper

Polymer Chemistry

Fig. 3 Conversion (■) and Đ (▼) vs. Eapp for eATRP of 50% (v/v) MA in [BMIm][OTf ]. Conditions: CMA : CCuBr2 : CTPMA : CRX = 552 : 0.1 : 0.1 : 1; t = 5 h.

agreement with Mn,th, and Đ = 1.10. Therefore, Eapp = E1/2 − 0.06 V was used in the successive reactions. Effect of temperature. An ionic liquid with a very high viscosity may impede mass transfer, especially when the reaction reaches high monomer conversion. The viscosity of ILs strongly depends on temperature, rapidly decreasing as the temperature is increased.46 The monomer propagation rate is also significantly affected by temperature. Therefore, eATRP of 50% (v/v) MA in [BMIm][OTf ] was performed at different temperatures from 25 °C to 75 °C. The results are shown in Table 2 and Fig. 4. Polymerization was well-controlled at all temperatures and, as expected, the reaction rate increased with T; conversion was almost quantitative after 1 h at 75 °C compared to 90% conversion in 5 h at 25 °C. The CuI catalyst, however, was a slightly weaker reducing agent in DMF at higher temperatures, in agreement with a previous report.37 The dispersity slightly increased with T, indicating the possible occurrence of side reactions or radical–radical termination. Nevertheless, very good polymerization control and molecular weight fidelity were retained up to temperatures as high as 75 °C. Comparison between [BMIm][OTf ] and CH3CN. The performance of the process in the ionic liquid was compared to eATRP in acetonitrile, performed under identical conditions, with the exception that all experiments in CH3CN were carried out in the presence of 0.1 M Et4NBF4 used as a supporting electrolyte. Virtually the same results were obtained in both solvents in terms of molecular weight, dispersity and the overall polymerization rate (see Table S1†). This confirms that [BMIm][OTf ] as a solvent for eATRP is as good as organic solvents, with the additional advantage that it does not require any supporting electrolyte.

Table 2

Fig. 4 (a) First-order kinetic plots and (b) evolution of Mn and Đ with conversion for eATRP of 50% (v/v) MA in [BMIm][OTf ] performed at 25 °C (triangles), 50 °C (squares) or 75 °C (circles). Conditions: CMA : CCuBr2 : CTPMA : CRX = 552 : 0.1 : 0.1 : 1; CCu = 10−3 M; Eapp = E1/2 − 0.06 V. The dotted line represents the theoretical Mn.

Galvanostatic eATRP. Galvanostatic electrolysis is in general more appealing than controlled-potential electrolysis because the required experimental setup is simpler: the reference electrode is not necessary, and a simple direct-current power supply can be used instead of a potentiostat/galvanostat. Therefore, eATRP of 50% (v/v) MA in [BMIm][OTf ] was carried out at 50 °C by applying a programmed constant current profile rather than a fixed potential. An appropriate profile, consisting of 4 current steps, was selected from the chronoamperometric curve recorded during a potentiostatic eATRP experiment performed under similar conditions (Fig. S2†). The two experiments yielded substantially the same results both in terms of the polymerization rate and properties of the produced polymer (Fig. S3†). Effect of DP and catalyst loading. Targeted DP was varied by changing the monomer amount from 25 to 75% (v/v, with respect to the total volume) at constant CRX. In all cases (Table 3, entries 1–3), reactions were fast and well-controlled (Fig. 5), with almost quantitative conversion in less than 2 h when 25% or 50% (v/v) of MA was used. The polymerization rate slightly decreased with increasing MA content, which is likely due to the diminished activation rate.43

eATRP of 50% (v/v) MA in [BMIm][OTf ] at different temperatures, Eapp = E1/2 − 0.06 Va app d

Entry

T (°C)

E1/2 b (V)

t (h)

Conversion (%)

Qc (C)

kp

Mn,th

Mn,app

Đ

1 2 3 4

25 50 75 50e

−0.58 −0.57 −0.50 −0.57

5.00 1.75 1.00 2.00

90 93 98 87

3.3 3.0 0.7 4.2

0.17 1.52 3.91 1.02

42 800 44 200 46 700 41 100

42 300 44 200 47 300 44 800

1.10 1.17 1.19 1.10

Conditions: CMA : CCuBr2 : CTPMA : CRX = 552 : 0.1 : 0.1 : 1. CCu = 10−3 M; Vtot = 5 mL. b vs. Fc+|Fc. c Total charge passed. d The slope of the ln([M]0/ [M]) vs. t plot. e Experiment at fixed current, with the following steps: 1.1 mA for 15 min, 0.95 mA for 17 min, 0.45 mA for 50 min, and 0.2 mA for 73 min.

a

650 | Polym. Chem., 2018, 9, 646–655


Polymer Chemistry Table 3

Paper

eATRP of MA in [BMIm][OTf ] at different targeted degrees of polymerization and catalyst loadings, Eapp = E1/2 − 0.06 V, T = 50 °Ca app c

Entry

DPtarget

MA/IL (v/v)

CCuII (mM)

t (h)

Conversion (%)

Qb (C)

kp

1 2 3 4 5

276 552 828 552d 552d

25/75 50/50 75/25 50/50 50/50

1.00 1.00 1.00 0.50 0.25

1.30 1.75 2.50 3.00 3.00

95 93 80 94 50

1.30 3.00 5.98 1.86 0.77

1.73 1.52 1.32 0.89 0.24

(h−1)

Mn,th

Mn,app

Đ

22 600 44 200 47 500 44 700 23 800

23 000 44 200 59 000 45 000 21 100

1.15 1.17 1.19 1.19 1.22

a

Conditions: CMA : CCuBr2 : CTPMA : CRX = x : 0.1 : 0.1 : 1, with x = 276 to 828. Vtot = 5 mL. b Total consumed charge. c The slope of the ln([M]0/[M]) vs. t plot. d Conditions: CMA : CCuBr2 : CTPMA : CRX = 552 : x : x : 1 with x = 0.05 or 0.025.

Fig. 5 (a) First-order kinetic plots and (b) evolution of Mn and Đ with conversion for eATRP of MA in [BMIm][OTf ] performed at Eapp = E1/2 − 0.06 V and T = 50 °C. Other conditions: CMA : CCuBr2 : CTPMA : CRX = x : 0.1 : 0.1 : 1, with x = 276 (squares), 552 (circles) or 828 (triangles); CCu = 10−3 M. The dotted lines represent the theoretical Mn.

Catalyst loading for the eATRP of MA in [BMIm][OTf ] (1 : 1, v/v) at 50 °C was reduced from 1 mM to 0.25 mM, which corresponds to variation of the copper/monomer molar ratio from 180 ppm to 45 ppm. Table 3 shows that polymerization was well-controlled with each catalyst loading. The polymerization rate increased with catalyst loading in agreement with a previous report.47 Fig. 6 reports the kinetic plots and the trends of Mn and Đ versus conversion, which were linear in all cases. Livingness of the polymerization: an electrochemical switch To verify the living character of the process, the applied potential was used to repetitively switch the system between active and dormant states. This was achieved by cycling Eapp between E1/2 − 0.06 V and E1/2 + 0.20 V. At Eapp = E1/2 − 0.06 V, polymerization was activated by converting CuII to CuI at the electrode. In contrast, at Eapp = E1/2 + 0.20 V, the activator was quenched by rapidly converting CuI to CuII and transforming the propagating radicals to dormant species (Scheme 1). Electrochemical oxidation of CuI quenched the polymerization much faster compared to other ATRP techniques, such as photo-ATRP,25 ARGET ATRP,48 or mechanically mediated ATRP.28 Indeed, in such polymerization methods the ON/OFF


Fig. 6 (a) First-order kinetic plots and (b) evolution of Mn and Đ vs. conversion for the eATRP of 50% (v/v) MA in [BMIm][OTf ]. Conditions: CMA : CCuBr2 : CTPMA : CRX = 552 : x : x : 1, x = 0.1 with CCu = 1 mM (squares), x = 0.05 with CCu = 0.5 mM (circles), and x = 0.025 with CCu = 0.25 mM (triangles). The dotted line represents the theoretical Mn.

cycle is realized by simply removing the external stimulus or the feeding of reducing agents, but polymerization continues until it is slowed down by radical–radical termination and consequent build-up of CuII species. Conversely, in eATRP a negligible monomer conversion was observed during the OFF periods, due to the fast generation of CuII, despite the high viscosity of ILs (Fig. 7a). The electrochemical ON/OFF switch was applied four times, increasing the monomer conversion from 0 to 19, 39, 58, and 77% during the active periods. Mn steadily increased in the ON periods, while no low Mn polymers were detected (Fig. 7b). Overall, the polymer chains grew as if there were no interruptions. These observations demonstrate the living character of the polymerization: efficient re-initiation of the chain ends resulted from minor termination and good preservation of the chain-end functionality. Retention of C–Br functionality: chain extension To further demonstrate the livingness of the polymerization, a poly(methyl acrylate) macroinitiator (PMA-Br) was extended with acrylonitrile (AN) to prepare a well-defined PMA-b-PAN copolymer. The PMA-Br macroinitiator was prepared under the same conditions as shown in Table 2, entry 2. Polymerization

Polym. Chem., 2018, 9, 646–655 | 651

Paper

Polymer Chemistry

Fig. 7 (a) Conversion vs. time plot and (b) evolution of Mn and Đ with conversion for the potentiostatic eATRP of 50% (v/v) MA in [BMIm][OTf ] under ON/OFF toggling. ON: Eapp = E1/2 − 0.06 V; OFF: Eapp = E1/2 + 0.20 V. The dotted line represents the theoretical Mn.

was stopped at ca. 50% conversion by applying Eapp = E1/2 + 0.20 V to rapidly convert the CuI activator to CuII. Then, the residual monomer was evaporated so that the remaining mixture contained IL, [Br–CuIITPMA]+, and a well-defined macroinitiator (PMA-Br) with Mn = 2.1 × 104 and Đ = 1.16. The complete removal of unreacted MA was confirmed by the NMR of the obtained polymerization mixture (Fig. S4†); the signals of the olefinic protons at 5.8–6.4 ppm are completely absent after monomer evaporation. Chain extension of PMA-Br with AN required a halogen exchange (HE), i.e. switching from a C–Br to a C–Cl chain end functionality (Scheme 2). The need for HE during chain extension is a consequence of a reactivity mismatch when crossing over from the PMA-Br chain end to the more active PAN-Br chain end.40 Propagation from the low fraction of initially formed PMA-b-PAN-Br is faster than initiation from the remaining PMA-Br, which results in low efficiency of crosspropagation. To solve this issue, the PAN-Br segment should be converted to a less reactive chain end. Both kact and KATRP for alkyl chloride-type of (macro) initiators are 1–2 orders of magnitude lower than those for alkyl bromides of the same structure. PAN-Cl is activated more slowly than PMA-Br,40 and thus Rp is decreased with respect to the rate of initiation from the initially added macroinitiator. This increases the initiation efficiency and enables preparation of a second block with a narrow dispersity.

Scheme 2

Mechanism of catalytic halogen exchange.

652 | Polym. Chem., 2018, 9, 646–655

Until now, preparation of block copolymers via HE was performed using high concentrations of the catalyst, i.e. adding an amount of CuICl/L equal to or higher than the amount of the Pn-Br chain end,41,42,49–56 which ensured complete conversion of Pn-Br to Pn-Cl. Conversely, HE was demonstrated to be inefficient under low-ppm ARGET ATRP57 or photo-ATRP58 conditions. Instead, we envisioned that a catalytic amount of the copper complex could efficiently convert Br to Cl in all Pn-Br chain ends, provided that enough Cl− was available. Therefore, we performed a catalytic halogen exchange (cHE) by simply adding an equimolar amount of tetraethylammonium chloride with respect to PMA-Br, exploiting the low-ppm, catalytic amount of CuII/L already present in the reaction mixture. The proposed mechanism of cHE is presented in Scheme 2. Once PMA-Br is activated (step I), the generated radical quickly adds one or more molecules of acrylonitrile, which is present in a large excess in the polymerization mixture (step II). In the presence of excess Cl−, most of the deactivator complex is converted to [Cl–CuIITPMA]+, because of the higher affinity of copper(II) for Cl− than Br− (step III)39,59,60. Then, the PAN chain end is preferentially deactivated by [Cl–CuIITPMA]+, regenerating the active [CuIL]+ and a Cl-capped dormant chain (step IV). Step IV is in principle in competition with a similar deactivation reaction involving [Br–CuIIL]+. Nevertheless, the halogen exchange reaction is strongly favoured by a much higher concentration of [Cl–CuIIL]+ than the bromide analogue. Indeed, we verified that addition of Cl− to a solution of [Br–CuIIL]+ immediately and quantitatively converts the latter to [Cl–CuIIL]+ (Fig. S5†). On the other hand, addition of excess Br− to a solution of [Cl–CuIIL]+ did not produce any [Br–CuIIL]+ (Fig. S6†). Therefore, the faster activation of Pn-Br than Pn-Cl in combination with the faster deactivation of P•n by [Cl–CuIIL]+ than [Br–CuIIL]+ ensured efficient halogen exchange, catalysed by a small amount of catalyst (1/10 with respect to the initial PMA-Br). eATRP of AN with the PMA-Br macroinitiator was performed at 50 °C in 50% (v/v) AN in [BMIm][OTf ] containing 10−2 M PMA-Br (CAN : CPMA-Br : CCuBr2 : CTPMA : CEt4NCl = 740 : 1 : 0.1 : 0.1 : 1.2). The catalyst concentration was 10−3 M and the applied potential was Eapp = E1/2 − 0.06 V. After 2 h of polymerisation, the copolymer started to precipitate and the reaction stopped. GPC analysis showed the formation of welldefined PMA-b-PAN-Cl with Mn,app = 2.9 × 104 and Đ = 1.11, with a clear shift of the molecular weight distribution to higher values (Fig. 8a). Conversion of acrylonitrile was determined to be 16% from 1H NMR analysis. This corresponds to a segment of PAN of DP = 118, with an overall Mn,th = 2.7 × 104 for the block copolymer, in good agreement with the Mn,app determined by GPC with PMMA calibration. Higher conversions of AN were not achieved because of the limited solubility of the copolymer in [BMIm][OTf ]/AN. 1H NMR (Fig. S7†) also showed the presence of PMA-b-PAN-Cl with ∼70% C–Cl functionality, while C–Br was not detected, suggesting the complete conversion of the chain-end functionality to the stronger C–Cl


Polymer Chemistry

Paper

Fig. 8 eATRP of 50% (v/v) AN in the [BMIm][OTf ] + PMA-Br macroinitiator performed at Eapp = E1/2 − 0.06 V; CAN : CPMA-Br : CCuBr2 : CTPMA : CEt4NCl = 740 : 1 : 0.1 : 0.1 : x: (a) x = 1.2; (b) x = 0. CCuBr2 = 10−3 M. GPC traces recorded before ( ) and after ( ) the chain extension.

bond. Overall, this result shows that C–Br was converted to C–Cl with a catalytic amount of the copper catalyst (1/10 with respect to the chain end). To further demonstrate the relevance and effectiveness of cHE, a PMA-Br macroinitiator was chain-extended without adding Et4NCl. In contrast to previous results, a bimodal GPC distribution was obtained (Fig. 8b), composed of the contribution of PMA-b-PAN-Br and unreacted PMA-Br. The faster activation of PMA-b-PAN-Br with respect to PMA-Br caused inefficient cross-propagation and poor control over macromolecular architecture, confirming that the catalytic halogen exchange was necessary and effective.

67 300, corresponding to the higher molecular weight fraction of the polymer PMA-Br (average molecular weight = 44 200). This finding suggests that if lower DPs are targeted, the polymer could be easily recovered by extraction as shorter polymers are more soluble in toluene. The recycling yield was 90%. After extraction of the polymer, the IL was reused for subsequent polymerizations. eATRP of 50% (v/v) MA in recycled [BMIm][OTf ] was comparable to the polymerization in fresh IL (Table 4). In recycled [BMIm][OTf ], polymerization was fast and controlled, with a first-order linear kinetic plot and a linear evolution of Mn as a function of conversion (Fig. 9). Actually, eATRP in the recycled IL was somewhat slower but with a lower dispersity. This can be due to the presence of excess bromide anions in the polymerization mixture, arising from radical–radical termination occurring during the previous polymerization; it is likely that Br− ions, present as [BMIm][Br], are not extracted by toluene. Excess bromide anions slow down the RX activation reaction,43,45 but, in some cases, can increase polymerization control.47 Further recycling showed results comparable to the first one (Table 4, entries 3 and 4), with a app slight and continuous decrease of kp , caused by accumu-

Recycling of the IL and catalyst eATRP in ILs can be considered an environmentally friendly process because it requires a low amount of catalyst that is continuously regenerated in the absence of chemical reducing agents. We also tested a procedure to recycle the catalyst in subsequent polymerizations. For this purpose, after a first polymerization, the residual monomer was evaporated and the polymer was extracted with toluene. Extraction was selective and almost quantitative because toluene is immiscible with [BMIm][OTf ]. The recovery of PMA-Br, after evaporation of toluene, was 98 mol% and no signals of toluene and [BMIm][OTf ] were detected by 1H NMR (Fig. S8†). In addition, [Br–CuIITPMA]+ quantitatively remained in the IL (Fig. S9†) because it is a very polar catalyst61 with low affinity for non-polar solvents such as toluene. GPC analysis of the recycled [BMIm][OTf ] showed the presence of a small amount of residual polymer (