European Polymer Journal 49 (2013) 4025–4034

Contents lists available at ScienceDirect

European Polymer Journal

journal homepage: www.elsevier .com/locate /europol j

Copolymerisation of e-caprolactone and trimethylene carbonatecatalysed by methanesulfonic acid

0014-3057/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.eurpolymj.2013.09.008

⇑ Corresponding authors. Tel.: +33 540 002 745 (F. Peruch), tel.: +35121 8417325 (M. Rosário Ribeiro).

E-mail addresses: rosario@ist.utl.pt (M.R. Ribeiro), peruch@enscbp.fr(F. Peruch).

João M. Campos a, M. Rosário Ribeiro a,⇑, M. Filipa Ribeiro a, Alain Deffieux b,c,Frédéric Peruch b,c,⇑a Instituto de Biotecnologia e Bioengenharia, Departamento de Engenharia Química, Instituto Superior Técnico, Universidade Técnica de Lisboa, Av. Rovisco Pais1, 1049-001 Lisboa, Portugalb Univ. Bordeaux, LCPO, UMR 5629, F-33600 Pessac, Francec CNRS, LCPO, UMR 5629, F-33600 Pessac, France

a r t i c l e i n f o

Article history:Received 13 June 2013Received in revised form 27 August 2013Accepted 9 September 2013Available online 25 September 2013

Keywords:CopolymerisationRing-opening polymerisationBiomaterialTrimethylene carbonatee-Caprolactone

a b s t r a c t

The copolymerisation of e-caprolactone (e-CL) and trimethylene carbonate (TMC) cata-lysed by methanesulfonic acid was investigated. Preliminary copolymerisation tests usinga monofunctional initiator confirm that the side bidirectional propagation previouslydetected in the homopolymerisation of TMC is also present in the copolymerisation. Thecomonomers in the e-CL/TMC system do not follow first order kinetics. The values of thereactivity ratios obtained by the Kellen–Tüdös method (re-CL = 2.90; rTMC = 0.62) suggestthat random copolymerisation can be achieved, although the copolymer will be richer ine-CL. Dihydroxyl-telechelic e-CL/ TMC random copolymers were prepared using a bifunc-tional initiator. 1H and 13C NMR, SEC and DSC measurements show that the poly(TMC-co-e-CL) samples presented the expected microstructural characteristics, a unimodal molar-mass distribution and a very narrow polydispersity. Based on these features a novel routefor the preparation of block copolyesters with tuned properties, and highly regarded in thedevelopment of materials for biomedicine, may be foreseen.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Synthetic aliphatic polyesters like poly(e-caprolactone),polylactide, polyglycolide (PCL, PLA, PGA) and relatedcopolymers are highly regarded in the development ofmaterials for biomedicine [1–3]. Aliphatic polycarbonateslike poly(trimethylene carbonate) (PTMC), on the otherhand, are not as attractive for biomedical applications,not only due to their weak mechanical properties, but alsodue to their higher stability in physiological conditions,strong hydrophobicity and lack of functionality, which re-duces their biological compatibility [4,5]. PTMC and relatedcopolymers found their place in the field of biomaterials as

components of block copolymers and in blends with brittlePLA, PCL or polyhydroxybutyrate (PHB) [5,6]. The copoly-merisation of trimethylene carbonate with cyclic esters of-fers a strategy to tune both the physicochemical propertiesand the degradation behaviour of the resulting materials.

Cationic homo and copolymerisation of e-caprolactoneand lactide monomers initiated by an alcohol and cata-lysed by trifluoromethanesulfonic acid (HOTf) or methane-sulfonic acid (MSA), proceed by an activated monomer(AM) mechanism in a controlled/living manner, withoutdetectable side processes [7–14].

The use of these acid catalysts is very interesting fromthe viewpoint of operational simplicity and environmentalcompatibility, particularly for MSA. Nevertheless, perform-ing the controlled polymerisation of TMC using such cata-lysts is rather problematic. This occurs because, in additionto the main AM mechanism, where the monomer activatedby the acid catalyst will undergo nucleophilic attack at the

4026 J.M. Campos et al. / European Polymer Journal 49 (2013) 4025–4034

carbonyl carbon atom by the initiating or propagating alco-hol (Scheme 1), there is a side process that reduces theaverage molar mass (as more than one chain of polymerper initiator molecule is formed) and broadens the molarmass distribution of the main process (Scheme 2) [15,16].

This side process was first proposed as resulting fromACE initiation (where the monomer activated by the acidcatalyst would ring-open, due to the nucleophilic attackof another non-activated monomer molecule) followedby spontaneous decarboxylation and subsequent com-bined (AM/ACE) bidirectional propagation, occurringrespectively at the hydroxyl and oxonium terminal chainmoieties, step (i) of Scheme 2. We recall here that ACEpropagation consists in a nucleophilic attack of an oxygenatom from the monomer on the a-carbon atom in tertiaryoxonium ion located at the growing chain end, whereasAM propagation involves a nucleophilic attack of an oxy-gen atom from the chain terminal OH group on the acti-vated monomer moiety. By keeping a low concentrationof TMC in the reactor (using multifeed/continuous feed ap-proaches), the active chain end initiation step (i) inScheme 2 can be suppressed, allowing polymerisation con-trol [15].

Very recently, it was shown that the combined (AM/ACE) bidirectional propagation shifts to a bidirectionalAM propagation (probably as consequence of an interme-diate proton shifting, step (ii) in Scheme 2), and globallythe propagation of this side process behaves as a bidirec-tional AM propagation. Accordingly, PTMC formed in thepresence of a monofunctional initiator, shows a bimodalmolar mass distribution and the polymer generated by sidebidirectional AM propagation always presented twice themolar mass of polymer originated from the main unidirec-tional AM propagation. It was also demonstrated that theuse of a bifunctional alcohol (leading to a situation whereall chain propagations are bidirectional) enables the prep-aration of PTMC presenting unimodal molar-mass distribu-tion and very narrow polydispersity, offering an alternativeway to prepare PTMC with a better controlled chain popu-lation [16].

This study aims to find if the proposed bifunctional ini-tiator approach can be successful for the preparation of hy-droxyl telechelic random poly(TMC-co-e-CL) withunimodal molar mass distribution using the environmen-tally friendly MSA catalyst. For this purpose, it was checkedin a preliminary study, if the proposed mechanistic path-ways developed for MSA-catalysed homopolymerisationof TMC still apply in the case of copolymerisation withe-caprolactone, and the reactivity ratios for the two mono-mers were determined, foreseeing the possibility toprepare random copolymers. Next, several copolymerisa-

Scheme 1. The Activated Momoner (AM) mechanism in the particular case of th(counter-anion omitted for clarity).

tions of e-CL and TMC were performed with 1,4-phenylen-edimethanol (PDM) as bifunctional initiator and thestructure of the copolymers obtained was investigated by1H and 13C NMR, SEC and DSC.

2. Experimental

2.1. Materials

All reactions and manipulations were performed underan inert atmosphere of argon, using standard Schlenk tech-niques. Commercial toluene (XiLab, 99%) and THF (JT Ba-ker, 99%) were first refluxed and distilled from CaH2 priorto use. Toluene was dried over polystyryl lithium and cryo-distilled. THF was dried over metallic sodium/benzophe-none and cryodistilled. Trimethylene carbonate (TMC,kindly donated by Boehringer) was dissolved in dry THF(0.7 g/mL) and stirred over CaH2 for 3 days, filtered, recrys-tallized twice from cold dry THF and finally dried undervacuum and stored at �20 �C in a flask under argon. Freshe-caprolactone (e-CL, Alfa Aesar, 99%) was stirred withCaH2 for two days, purified by fractionated distillation atreduced pressure and stored under argon. Biphenyl-4-methanol (BPM, Alfa Aesar, 98%) was recrystallized fromdichloromethane and 1,4-phenylenedimethanol (PDM,Roth, 99%) was used as received. The internal standardused for 1H NMR, 1,2-diphenylethane (DPE) was recrystal-lized from petroleum ether, dried under vacuum andstored in a flask under argon. Methanesulfonic acid (MSA,Aldrich, 99.5%) was bubbled for some hours with argon,prior to use, then stored in a flask under this gas. N,N-diiso-propylethylamine (DIPEA, Aldrich, 99%) was used asreceived.

2.2. Polymerisation procedure

Preliminary e-CL/TMC copolymerisations were per-formed in order to investigate the mechanistic featuresand calculate the reactivity ratios. Monomer feed ratiosof 1:4, 1:1 and 4:1 were used, keeping the total initialmonomer concentration [e-CL]0 + [TMC]0 at 1 M. For thetest with [e-CL]0 = [TMC]0, trimethylene carbonate (1.1 g,10.8 mmol), the monofunctional initiator biphenyl-4-methanol (22.0 mg, 0.117 mmol, 0.117 mmol OH, �0.005equivalents of total monomer) and the NMR internal stan-dard 1,2-diphenylethane (200 mg, 1.1 mmol, �0.05 equiv-alents of total monomer) were introduced under argon in a100 ml round-bottom flask. e-CL (1.2 mL, 10.8 mmol) anddry toluene (20 mL) were added, under argon, from bur-ettes. The flask was placed in an oil bath set to 30 �C and

e ring-opening polymerisation of TMC catalysed by methanesulfonic acid

Scheme 2. The side reaction present for the ring-opening polymerisation of TMC catalysed by methanesulfonic acid (counter-anion omitted for clarity).

J.M. Campos et al. / European Polymer Journal 49 (2013) 4025–4034 4027

after homogenisation the methanesulfonic acid catalyst(8 ll, 0.123 mmol, �1 equivalent of initiator) was addedwith a microsyringe under argon to start the polymerisa-tion. Small aliquots were taken to check monomer conver-sions by 1H NMR spectroscopy (each 10–20 min in the firsttwo hours of reaction), using deuterated solvent with asmall amount of DIPEA (Hünig’s base) to quench the MSAcatalyst and stop the polymerisation during analysis. Thecopolymerisation was surveyed by this method for 7 h.At the end the polymer was recovered by precipitation incold methanol and washed with cold pentane (or petro-leum ether) in turn with additional cold methanol, andthoroughly dried under vacuum before SEC analysis. Theprocedure was further repeated with [e-CL]0/[TMC]0 ratiosat 1:4 and 4:1, with the monomer amounts being adjustedaccordingly. The consumption data for each comonomerwas plotted as a function of time. Second order polynomi-als were fitted to the consumption curves of each mono-mer. The copolymer composition at the beginning of thecopolymerisation was calculated from the slopes of the fit-ted curves at the origin of time [17]. The copolymer com-position data for the three copolymerisations was thenused for the calculation of reactivity ratios, by the Kel-len–Tüdös method [18].

For the preparation of dihydroxyl-telechelic poly(TMC-co-e-CL) with unimodal molar mass distributions,

the polymerisation procedure used is basically the same,but making use of a bifunctional alcohol initiator. Thecopolymerisation was performed with [e-CL]0 = [-TMC]0 = 0.5 M and using 1,4-phenylenedimethanol as initi-ator (�0.005 equivalents of total monomer). Monomerconversions were again surveyed by 1H NMR spectroscopyin deuterated solvent containing DIPEA. The reaction wasintentionally stopped before the full conversion of themore reactive monomer e-CL (�80–90%), in order to avoidthe formation of terminal blocks of PTMC. Finally, theproduct poly(TMC-co-e-CL) was recovered as describedabove, and characterised by 13C NMR, 1H NMR, SEC andDSC.

2.3. Measurements and characterisation

1H and 13C NMR spectra were recorded in CDCl3 at 25 �Con a Bruker Avance 400 equipment, at 400 and 100 MHz,respectively. For the copolymerisations of e-CL with TMC,the monomer conversions were calculated by 1H NMRusing the peak areas of the methylene group in the a-posi-tion to the carbonyl in e-CL (CH2C(O), d = 2.63 ppm) andthe side methylene groups in TMC (CH2CH2CH2,d = 4.45 ppm), with the peak area for the signal of the ben-zylic protons of DPE (CH2CH2, d = 2.93 ppm) as internalstandard (see ESI, Fig. S1). The fractions of each monomer

4028 J.M. Campos et al. / European Polymer Journal 49 (2013) 4025–4034

in precipitated copolymer samples (Fe-CL and FTMC) werecalculated from the peak areas for the central methylenegroups in e-CL units (CH2CH2CH2CH2CH2, d = 1.39 ppm)and in TMC units (CH2CH2CH2, d = 2.02 ppm).

The number-average molar masses and the molar-massdispersities of the polymeric samples were measured bysize exclusion chromatography (SEC) at 40 �C with a Var-ian/Polymer Laboratories PL-GPC 50 Plus chromatographequipped with RI and UV detectors, with TosohG4000HXL, G3000HXL and G2000HXL columns, calibratedwith polystyrene standards. Tetrahydrofuran (THF) wasused as eluent with a flow rate of 1.0 mL/min. For the e-CL/TMC copolymer, the estimated number-average molarmasses, Mn, are calculated from the apparent molar massesgiven by SEC, MnSEC, the fractions of e-CL and TMC in thecopolymer (Fe-CL and FTMC, as determined from 1H NMR)and the correction factors for these monomers (0.73 forTMC and 0.56 for e-CL as reported previously) [19]:

Mn ¼ MnSECð0:56Fe-CL þ 0:73FTMCÞ

Bimodal RI SEC profiles were split in its two componentpeaks using model chromatography functions (exponentiallymodified Gaussians) and the deconvolution proceduresavailable in the PeakFit™ software. The corresponding Mn

and Mw/Mn values were then evaluated, with the value ofMn being corrected with the factors mentioned above.

The thermal properties of the synthesised materialswere measured by differential scanning calorimetry(DSC), using a TA Instruments Q100 equipment. The poly-meric samples (�10–15 mg) were enclosed in aluminiumcapsules and analysed in the temperature range from�100 �C to 100 �C, with an heating rate of 10 �C/min, undera flow of helium (25 mL/min). The data presented refer tothe second heating scan.

3. Results and discussion

3.1. Copolymerisation mechanism features and determinationof reactivity ratios

For the present investigation, the copolymerisations ofTMC with e-CL initiated by monofunctional biphenyl-4-methanol (BPM) and catalysed by methanesulfonic acid(MSA) were performed as a preliminary study, to check ifthe mechanistic pathway proposed for TMC homopolymer-isations is also applicable in copolymerisation conditions,

Scheme 3. Preliminary copolymerisations of TMC and e-CL init

according to Scheme 3. This study will also provide datato evaluate the kinetics of the copolymerisation and calcu-late the reactivity ratios for the e-CL/TMC system in thechosen experimental conditions.

The main results of the three copolymerisation experi-ments are summarised in Table 1. The measured molarmasses are inferior to the calculated ones, as expected.Due to the side ACE initiation from TMC moieties, morethan one chain of polymer per initiator molecule is formed.Moreover, even if the molar mass dispersities appeared tobe low, one can see in Fig. 1 that bimodal distributionswere clearly present for the SEC profiles obtained withrefractive index (RI) detector. On the contrary, SEC profilesobtained with the UV detector (detecting selectively chainsinitiated by aromatic monofunctional BPM, via AM initia-tion) are monomodal. As shown recently, these observa-tions confirm the existence of a side bidirectionalpropagation process initiated from TMC (ACE initiation),generating additional chains with higher molar mass[15]. These peaks in the RI SEC profiles were deconvolutedusing model functions derived from the narrow unimodalpeaks obtained with the UV detector, and the correspond-ing Mn and Mw/Mn were calculated [16]. The results arepresented in Fig. 1 and rightmost columns in Table 1.

The deconvoluted SEC profiles show that once again thechain populations resulting from the side ACE initiationfollowed by bidirectional propagation (Scheme 3) and fromthe main monodirectional AM propagation (Scheme 3)present a ratio molar mass of 2:1. These measurementssupport the earlier claim that the combined AM/ACE prop-agation shifts to bidirectional AM propagation. In fact, in acombined AM/ACE propagation this 2:1 M ratio will bequite unlikely, as most probably AM and ACE will presentquite different rates for monomer consumption, and sidereactions, like crosslinking and transesterification, areprone to occur for ACE propagation. This way we may con-clude that the mechanistic pathways for the polymerisa-tion of TMC catalysed by MSA also apply for itscopolymerisation with e-CL.

The compositions of the copolymer at the beginning ofthe reaction (as molar fractions Fe-CL, FTMC) were obtainedfrom the slopes of the consumption curves at the originof time (see ESI, Figs. S2–S4), as these represent the conver-sion of each monomer at the beginning of the copolymer-isation [17]. The Kellen–Tüdös method was chosen for thecalculation of reactivity ratios because it provides a betterdistribution of data points over the abscissa scale [18].

iated by a monoalcohol, and its main and side reactions.

Table 1Preliminary copolymerisations using the e-CL/TMC system: experimental conditions and results of characterisation.

Run [e-CL]0

(M)[TMC]0

(M)[MSA]/[OH]b

Time(h)

xe-CLc xTMC

c Fe-

CLd

Mnthe

(g mol�1)RI Data UV Data Deconvoluted RI Datah

Mnf

(g mol�1)Mw/Mn

gMn

f

(g mol�1)Mw/Mn

gMn

f

(g mol�1)Mw/Mn

gFACE

i

R1a 0.20 0.83 1.0 7.7 0.35 0.18 – – – – – – – – –22.8 �1.0 0.60 0.29 10,710 9380 1.18 7010 1.05 15,310/

76901.04/1.04

0.32

R2 0.51 0.51 1.1 7.1 0.50 0.31 0.66 8340 5660 1.15 5071 1.05 10,000/5100

1.08/1.07

0.11

R3 0.82 0.19 1.1 7.0 0.54 0.33 0.88 12,995 7140 1.09 6727 1.04 12,530/6880

1.08/1.04

0.05

Polymerisations carried out in toluene at 30 �C, using biphenyl-4-methanol (BPM) as initiator and methanesulfonic acid (MSA) as catalyst, with [TMC]0 + [e-CL]0 = 1.0 M.

a After 7 h of kinetic measurements, experiment R1 was left reacting for a total of 22.8 h before recovering and analysing the product.b [OH] = [BPM]0 = 5.5 � 10�3 M for monofunctional initiator BPM.c Conversions calculated by 1H NMR spectroscopy.d Composition of the recovered polymer (molar fraction), as determined by 1H NMR (see Experimental).e Theoretical number-average molar mass Mnth = [TMC]0/[BPM]0 �MTMC � xTMC + [e-CL]0/[BPM]0 �Me-CL � xe-CL + MBPM.f Corrected number-average molar mass Mn for poly(TMC-co-e-CL) calculated as Mn = MnSEC � (0.56 Fe-CL + 0.73 FTMC) with MnSEC being the apparent

molar mass.g Molar mass dispersities estimated from raw SEC data.h Deconvolution performed with EMG functions.i Fraction of chains initiated via ACE and propagating by combined AM/ACE, estimated from the areas under the deconvoluted RI SEC peaks.

Fig. 1. SEC profiles obtained for the products of the preliminary e-CL/TMCcopolymerisations, samples R1–R3 (Table 1). Traces for two detectors,refractive index (upper plot) and UV–Vis (lower plot). Deconvolution ofthe RI traces is also shown (red/blue lines), identifying the populationscoming from main and side reactions. (For interpretation of the referencesto colour in this figure legend, the reader is referred to the web version ofthis article.)

J.M. Campos et al. / European Polymer Journal 49 (2013) 4025–4034 4029

These were obtained from parameters evaluated by linearregression (Fig. 2 and Table S1) and their values are:

Re-CL ¼ 2:90� 0:14; rTMC ¼ 0:62� 0:08

The errors were derived from the standard deviations inthe slope and intercept obtained in the regression. Thesereactivity ratio values, with re-CL > 1 and rTMC < 1, indicate

that the polymerisation of e-CL is favoured and that thepolymer initially formed will contain little TMC. Neverthe-less, the comonomers are inserted essentially in a randommanner [20]. The copolymer is therefore expected to con-tain a larger proportion of e-CL, but the two monomers willbe in a random placement.

Due to the existence of a side propagation, the signifi-cance of a calculation of reactivity ratios in the case ofthe copolymerisation of TMC and e-CL could be ques-tioned. Here, it is important to notice that the mechanismshift, occurring in the side reaction (steps (ii)–(iii) inScheme 2) implies that all propagations proceed by thesame AM mechanism, regardless of the way they are initi-ated (AM or ACE initiation). Moreover, the SEC results dis-cussed above suggested that e-CL did not interfere withthis mechanistics. The two comonomers are depleted bythe same AM process, be it in the main or side reactions.The reactivity ratios are therefore meaningful, and referto the AM propagation process, that is, the insertion ofTMC and e-CL monomers in hydroxyl-terminated chains.

3.2. Preparation of dihydroxyl-telechelic poly(TMC-co-e-CL)from a bifunctional initiator

Next, several copolymerisations of e-CL and TMC wereperformed with 1,4-phenylenedimethanol (PDM) asbifunctional initiator. According to Scheme 4, this processwill lead to a mixture of dihydroxyl-telechelic poly(TMC-co-e-CL) chains, differing only in the central unit, but form-ing an uniform population with the same molar mass. Themain results are summarised in Table 2.

Fig. 3 displays the 1H NMR spectrum for a representa-tive poly(TMC-co-e-CL) sample (sample M1, Table 2). Thetwo singlet peaks at d = 5.12 and 5.16 ppm can be recogni-sed as the benzylic protons present in the initiator groupsof PDM [16]. The one at higher chemical shift is easilyascribed to structures where TMC was the first monomer

Fig. 2. Copolymerisation curve (left) and Kellen–Tüdös regression (right) for determination of reactivity ratios for the e-CL/TMC system. The data used canbe found in Table S1.

Scheme 4. Copolymerisations of TMC and e-CL initiated by a diol, and its main and side reactions.

Table 2Preparation and characterisation of poly(TMC-co-e-CL) dihydroxyl-telechelic copolymers and data for PTMC and PCL homopolymers.

Run Time (h) xe-CLa xTMC

a Mnthb (g mol�1) Fe-CL

c Mnd (g mol�1) Mw/Mn

e Tg (�C) Tm (�C) DHm (J/g) wcf (%)

M1 23.5 0.91 0.67 30,900 0.56 24,860 1.10 �51 – – –M2 23 0.82 0.55 30,000 0.58 22,845 1.10 g g g g

M3 23 0.80 0.51 27,500 0.59 22,930 1.10 �51 – – –PCL 1.5 1.0 0.0 10,400 1.0 10,300 1.04 �60 55 86 62PTMC 11 0.0 1.0 15,400 0.0 14,800 1.07 �29 – – –

Polymerisations carried out in toluene at 30 �C, using PDM as bifunctional initiator and methanesulfonic acid (MSA) as catalyst, with [TMC]0 = [e-CL]0 = 0.5 M, [PDM]0 = 2.7 � 10�4. [OH] = 2[PDM]0 for PDM, [MSA]/[OH] = 1.1.

a Conversions calculated by 1H NMR.b Theoretical number-average molar mass, Mnth = [TMC]0/[PDM]0 �MTMC � xTMC + [e-CL]0/[PDM]0 �Me-CL � xe-CL + MPDM.c Composition of the recovered polymer (molar fraction), as determined by 1H NMR.d Corrected number-average molar mass Mn for poly(TMC-co-e-CL) calculated as Mn = MnSEC � (0.56 Fe-CL + 0.73 FTMC) with MnSEC being the apparent

molar mass.e Molar-mass dispersities Mw/Mn estimated from SEC data.f Mass fraction of crystallinity, calculated using as reference the heat of fusion of 100% crystalline PCL, DH0,m = 139.4 J/g. [27].g Not determined.

4030 J.M. Campos et al. / European Polymer Journal 49 (2013) 4025–4034

linked to a CH2OH group of the initiator, while the other atlower chemical shift represents the same structure withe-CL as first monomer. A comparison of the respective peakareas shows that the proportion of chains initiated withe-CL is around 68%, demonstrating again the higher reac-tivity of this monomer towards the chosen initiator. Theabsence of any neighbouring peaks in this region suggeststhat no free benzylic CH2OH groups remain, therefore

implying that a bidirectional chain propagation was ob-tained from the PDM initiator, as intended. Unfortunately,chains originated from the side propagation present a cen-tral trimethylene unit and cannot be distinguished becauseall monomer segments present similar sequences of multi-ple methylene units. The two multiplet peaks at d = 3.65and 3.73 ppm represent the terminal methylene protonslocated at the chain ends. Again, the one at higher chemical

Fig. 3. 1H NMR spectrum (CDCl3, 400 MHz) for poly(TMC-co-e-CL) sample M1.⁄: acetone.

J.M. Campos et al. / European Polymer Journal 49 (2013) 4025–4034 4031

shift is ascribed to TMC and the other one to e-CL [14–16].The relative peak areas for these signals show that the lastmonomer inserted was TMC for 74% of the chain ends.Since the analysis of chain ends (see ESI, Fig. S5) and thereactivity ratios both indicate that e-CL is more reactivethan TMC, then in a closed batch reactor the concentrationof the former will decrease faster. As the copolymerisationproceeds, the proportion of TMC will become higher, there-fore favouring its insertion at the end of the chains,explaining the observed result. The region betweend = 4.0 and 4.3 ppm shows the multiplet peaks resultingfrom methylene protons in the CH2-ester and CH2-carbon-ate combinations. These represent four possible se-quences/diads for monomer addition, in order ofincreasing chemical shift: e-CL–e-CL, TMC–e-CL, e-CL–TMC and TMC–TMC. These peak assignments were madeaccording to the literature [21,22]. It should be noticed thatin the AM mechanism the ring-opened monomers are

Fig. 4. 13C NMR spectrum (CDCl3, 100 MHz

inserted through the carbonyl, leaving the hydroxyl groupat the end of the chain, while in the coordination-insertionmechanism this process is inverted. The net result is thatthe diad heterosequences must be switched, while thepeak chemical shifts are the same for the copolymers ob-tained from each mechanism. The relative frequency foreach diad was estimated by approximate integration ofthe multiplets. Together, the heterosequences TMC–e-CLand e-CL–TMC make for 50% of the diads, while e-CL–e-CL and TMC–TMC amount to 32% and 18%, respectively.This result suggests the formation of random copolymerswithout large sequences of a same monomer in the chains.

Fig. 4 presents the 13C NMR spectrum for the samecopolymer sample M1. The overall peak assignment wasmade according to data reported in the literature for thesame type of random copolymer and for poly(TMC) andpoly(e-CL) homopolymers prepared using organometalliccatalysts, again taking into account the inverted insertion

CH2OH

) for poly(TMC-co-e-CL) sample M1.

Fig. 5. SEC profiles for poly(TMC-co-e-CL) samples M1 and M2. Traces fortwo detectors, refractive index (upper plot) and UV–Vis (lower plot).

4032 J.M. Campos et al. / European Polymer Journal 49 (2013) 4025–4034

of the monomers [19,22–24]. The characteristic peaks forthe two monomers show a heavy splitting into multiplewell-defined signals. This is clearly visible for the peaks as-cribed to carbons in a position to oxygen, at d = 61.5–68.0 ppm, where the splitting of the main signals of thee-CL–e-CL and TMC–TMC homosequences is due to theexistence of TMC–e-CL and e-CL–TMC heterosequences[22,24]. A similar effect is observed for the carbonyl peakscentred at d = 155.1 and 173.5 ppm, ascribed to TMC and e-CL chain units, respectively. Each one of these carbonylpeaks is split twice, into two doublets with comparableintensities, a feature which confirms the presence of theeight sequences of TMC and e-CL carbonyl triads in similaramounts [23]. These results further suggest that the poly-mer chains are mostly composed by random sequencesof TMC and e-CL.

Overall, the 1H and 13C NMR data confirms the trends tobe expected from the determined reactivity ratios. Whilethese suggest that a random e-CL–TMC copolymer will beobtained, in a system without continuous feed like theone used in the present work the more reactive e-CL ispreferentially inserted at the beginning of the polymerisa-tion and will be depleted faster. As the polymerisation pro-ceeds, the concentration of unreacted TMC rises and itsinsertion in the chain becomes increasingly favourable. Agradient copolymer should be obtained, richer in e-CL inthe beginning of the polymerisation, and richer in TMC asthe propagation proceeds. Since the polymerisation wasinterrupted before complete depletion of e-CL, no terminalPTMC blocks are formed and the last monomer unit can beeither e-CL or TMC, but the probability of being TMC ishigher. For bidirectional propagation from a diol initiatorlike PDM, this means that the chains are richer in e-CLaround the centre and richer in TMC towards the ends.Nevertheless the local composition through the chainsshould be completely random.

Fig. 5 shows the typical SEC profiles obtained with RIand UV detectors for a poly(TMC-co-e-CL) sample initiatedby the bifunctional alcohol. Contrarily to the observed fea-tures of the samples generated from monofunctional BPMinitiator, the molar-mass distributions determined for allsamples are now clearly unimodal, with narrow molarmass dispersities, Mw/Mn � 1.10 (Table 2). The minorpeak/shoulder with estimated molar mass halved rela-tively to the main peak (for M1 the peak molar masses,Mp, were 27,000 and 13,300 g mol�1) is much probablycaused by the formation of some extra copolymer frommonofunctional impurities present in the initial system,probably water. If it can be assumed that this small peakcomes from a stray unidirectional propagation, then itslocation in the RI profile and its absence in the UV–Visone confirm that the main peak is due to a copolymerformed by bidirectional propagation from PDM initiator,as intended [16]. The molar masses of the copolymer,measured by SEC, were corrected applying the correctionfactors which take into account the determined composi-tion of the copolymer and the correction factors for eachmonomer (see Section 2). The estimated Mn is lower thanthe theoretical Mn. This should be ascribed mainly to thepresence of the unaccounted propagations generated bycontaminant traces and by the side propagation

mechanism with decarboxylation of TMC monomer[15,16]. But it is also important to notice that the correc-tion of the raw values of Mn by the method of correctionfactors described above is already a rough approximation.

3.3. Characterisation by DSC

To further characterise the copolymers, their thermalproperties were measured by DSC. The melting and glasstransition temperatures for PCL are located around 55 �Cand at �60 �C, respectively, depending on its molar massand crystallinity. On the other hand, PTMC is amorphousand presents a Tg which has been reported around �15to �20 �C [5,6,25]. The thermal characteristics ofpoly(TMC-co-e-CL) copolymers (samples M1 and M3) wereevaluated by DSC and compared with those of some PCLand PTMC homopolymers also prepared using MSA andPDM (Fig. 6 and Table 2). For the copolymers, the analysisshowed the absence of any melting peak and detected asingle glass transition at Tg = �51 �C. For statistical copoly-mers obtained from two monomers, the Fox equation re-lates the Tg of the copolymer and of the twohomopolymers as:

1=Tg ¼ m1=Tg;1 þm2=Tg;2

where m1, m2 are the mass fractions of the two comono-mers and Tg,1, Tg,2 the glass transition temperatures of thecorresponding homopolymers [26]. This equation predictsa Tg of �48 and �49 �C for M1 and M3, respectively, whichis not far from the measured result. The crystallinity of PCLwas therefore completely suppressed in the copolymerisa-tion of e-CL with TMC, while the glass transition tempera-ture of the product obtained is compatible with theformation of a random copolymer.

Fig. 6. DSC thermograms (second heating scan) obtained for poly(TMC-co-e-CL) samples M1 and M3, compared to PCL and PTMC homopolymersamples. The inset shows the detail of the Tg region for PCL.

J.M. Campos et al. / European Polymer Journal 49 (2013) 4025–4034 4033

4. Conclusions

This work proposes a method to obtain dihydroxyl-telechelic poly(TMC-co-e-CL) copolymers by acid-cata-lysed ring-opening polymerisation of cyclic monomers ini-tiated by a bifunctional initiator/diol, which is the keyfeature in this strategy. Bidirectional propagation solvesthe known problem of controlling the populations ofTMC chains, and allows a faster chain growth, given thesimultaneous propagation by both ends instead of justone. The exploratory copolymerisations made with themonofunctional initiator showed that the ratio of molarmasses for copolymer chains originated by the side AM/ACE-initiated propagation and for the main AM propaga-tion keep the ratio of 2:1, supporting the claim of fastmechanism-shifting from AM/ACE to bidirectional AM,made in our previous investigation. This warrants thatthe use of a bifunctional initiator in the copolymerisationof e-CL with TMC will lead to narrow unimodal copoly-mers. The determination of the reactivity ratios showedthat the copolymerisation of e-CL and TMC leads to ran-dom copolymers where e-CL is inserted preferentially. Ina closed (batch) system, a gradient random copolymershould be obtained. The results of NMR, SEC and DSC char-acterisation showed that the poly(TMC-co-e-CL) dihydr-oxyl-telechelic copolymers prepared in this work presentthe expected structural features. The present study con-firms and complements a previous investigation, andopens a new and promising route for the preparation ofblock copolymers, with tuned properties, from TMC. Theunimodal molar-mass distribution and very narrow poly-dispersity of telechelic random poly(TMC-co-e-CL) makesit a very suitable candidate to be used as macroinitiatorfor the sequential polymerisation with other cyclic mono-mers. Additionally, the tuning of the properties of theresulting materials can be performed by an appropriateselection of the length and type of each block used, softor rigid. This way the preparation of block copolyesters, de-rived from lactones and cyclic carbonates, with tunedproperties and free of residues, may be envisaged for bio-medical applications (such as biodegradable elastomersin flexible implants, drug delivery matrices and tissuescaffolds).

Acknowledgements

The authors would like to thank Boehringer for the giftof TMC. JMC thanks the Fundação Para a Ciência e a Tecn-ologia for his post-doctoral research grant (SFRH/BPD/63275/2009).

Appendix A. Supplementary material

Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.eurpolymj.2013.09.008.

References

[1] Sisson AL, Schroeter M, Lendlein A. Polyesters. In: Lendlein A, SissonAL, editors. Handbook of biodegradable polymers. Wiley-VCH VerlagGmbH & Co. KGaA; 2011. p. 1–21.

[2] Domb AJ, Kumar N, Ezra A. Biodegradable polymers in clinical useand clinical development. Hoboken, New Jersey: John Wiley & Sons,Inc.; 2011.

[3] Woodruff MA, Hutmacher DW. The return of a forgotten polymer–polycaprolactone in the 21st century. Prog Polym Sci2010;35(10):1217–56.

[4] Feng J, Zhuo RX, Zhang XZ. Construction of functional aliphaticpolycarbonates for biomedical applications. Prog Polym Sci2012;37(2):211–36.

[5] Suriano F, Coulembier O, Hedrick JL, Dubois P. Functionalized cycliccarbonates: from synthesis and metal-free catalyzed ring-openingpolymerization to applications. Polym Chem 2011;2(3):528–33.

[6] Engelberg I, Kohn J. Physico-mechanical properties of degradablepolymers used in medical applications: a comparative study.Biomaterials 1991;12(3):292–304.

[7] Kubisa P, Penczek S. Cationic activated monomer polymerization ofheterocyclic monomers. Prog Polym Sci 1999;24(10):1409–37.

[8] Basko M, Kubisa P. Cationic copolymerization of epsilon-caprolactone and L,L-lactide by an activated monomer mechanism.J Polym Sci A Polym Chem 2006;44(24):7071–81.

[9] Basko M, Kubisa P. Polyester oligodiols by cationic AMcopolymerization of L,L-lactide and epsilon-caprolactone initiatedby diols. J Polym Sci A Polym Chem 2007;45(14):3090–7.

[10] Basko M, Kubisa P. Mechanism of propagation in the cationicpolymerization of L,L-Lactide. J Polym Sci A Polym Chem2008;46(23):7919–23.

[11] Basko M, Kubisa P. Cationic polymerization of L,L-Lactide. J Polym SciA Polym Chem 2010;48(12):2650–8.

[12] Bourissou D, Martin-Vaca B, Dumitrescu A, Graullier M, Lacombe F.Controlled cationic polymerization of lactide. Macromolecules2005;38(24):9993–8.

[13] Bourissou D, Susperregui N, Delcroix D, Martin-Vaca B, Maron L.Ring-opening polymerization of epsilon-caprolactone catalyzed bysulfonic acids: computational evidence for bifunctional activation. JOrg Chem 2010;75(19):6581–7.

[14] Gazeau-Bureau S, Delcroix D, Martin-Vaca B, Bourissou D, Navarro C,Magnet S. Organo-catalyzed ROP of epsilon-caprolactone:methanesulfonic acid competes with trifluoromethanesulfonicacid. Macromolecules 2008;41(11):3782–4.

[15] Delcroix D, Martin-Vaca B, Bourissou D, Navarro C. Ring-openingpolymerization of trimethylene carbonate catalyzed bymethanesulfonic acid: activated monomer versus active chain endmechanisms. Macromolecules 2010;43(21):8828–35.

[16] Campos JM, Ribeiro MR, Ribeiro MF, Deffieux A, Peruch F. A newinsight into the mechanism of the ring opening polymerization oftrimethylene carbonate catalysed by methanesulfonic acid.Macromol Chem Phys 2013;214(1):85–93.

[17] Bauduin G, Boutevin B, Belbachir M, Meghabar R. Determination ofreactivity ratios in radical copolymerization – a comparison ofmethods for a methacrylate n-vinylpyrrolidone system.Macromolecules 1995;28(6):1750–3.

[18] Kelen T, Tüd}os F. A new improved linear graphical method fordeterming copolymerization reactivity ratios. Reaction KineticsCatal Lett 1974;1(4):487–92.

[19] Palard I, Schappacher M, Belloncle B, Soum A, Guillaume SM.Unprecedented polymerization of trimethylene carbonate initiatedby a samarium borohydride complex: mechanistic insights and

4034 J.M. Campos et al. / European Polymer Journal 49 (2013) 4025–4034

copolymerization with epsilon-caprolactone. Chem Eur J2007;13(5):1511–21.

[20] Gnanou Y, Fontanille M. Organic and physical chemistry ofpolymers. Hoboken, New Jersey: John Wiley & Sons, Inc.; 2008.

[21] Ling J, Zhu W, Shen Z. Controlling ring-opening copolymerization ofe-caprolactone with trimethylene carbonate by scandium Tris(2,6-di-tert-butyl-4-methylphenolate). Macromolecules 2004;37(3):758–63.

[22] Zhao B, Lu CR, Shen Q. Ring-opening polymerization oftrimethylenecarbonate and its copolymerization with epsilon-caprolactone by lanthanide(II) aryloxide complexes. J Appl PolymSci 2007;106(2):1383–9.

[23] Agarwal S, Naumann N, Xie X. Synthesis and microstructuralcharacterization of ethylene carbonate–e-caprolactone/L-lactidecopolymers using one- and two-dimensional NMR spectroscopy.Macromolecules 2002;35(20):7713–7.

[24] Schappacher M, Fabre T, Mingotaud AF, Soum A. Study of a(trimethylenecarbonate-co-epsilon-caprolactone) polymer – Part 1:preparation of a new nerve guide through controlled randomcopolymerization using rare earth catalysts. Biomaterials2001;22(21):2849–55.

[25] Pego AP, Poot AA, Grijpma DW, Feijen J. Copolymers of trimethylenecarbonate and epsilon-caprolactone for porous nerve guides:synthesis and properties. J Biomater Sci – Polym Ed 2001;12(1):35–53.

[26] Sperling LH. Introduction to physical polymer science. Hoboken,New Jersey: John Wiley & Sons, Inc.; 2006.

[27] Crescenzi V, Manzini G, Calzolari G, Borri C. Thermodynamics offusion of poly-b-propiolactone and poly-e-caprolactone. Comparativeanalysis of the melting of aliphatic polylactone and polyester chains.Eur Polym J 1972;8(3):449–63.