DOI: 10.1002/chem.200802388

Radical Polymerization of Vinyl Acetate withBis(tetramethylheptadionato)cobalt(II): Coexistence of Three Different

Mechanisms

K. S. Santhosh Kumar,[b] Yves Gnanou,*[b] Yohan Champouret,[a]

Jean-Claude Daran,[a] and Rinaldo Poli*[a, c]

Introduction

In recent years, rapidly increasing attention has been paid tocontrolled radical polymerization (CRP).[1–6] Consequently,novel polymeric materials with well-defined molecular pa-rameters such as average molecular weight, molecular-weight distribution, chain-end functionality, and topologicalstructure have become accessible through various strategies.A current challenge of CRP is to control the radical poly-merization of monomers that generate more reactive radi-cals such as vinyl acetate, vinyl chloride, vinylidene chloride,and ethylene.

A variety of methods have been devoted to mediating thepolymerization of vinyl acetate (VAc), such as degenerativechain transfer with alkyl iodides,[7] dithiocarbamates andxanthates (RAFT/MADIX),[8–10] and atom-transfer radicalpolymerization.[11,12] The use of bis(b-diketonato)cobalt(II)

Abstract: The complex [CoII ACHTUNGTRENNUNG(tmhd)2](4 ; tmhd= 2,2,6,6-tetramethylhepta-3,5-dionato) has been investigated as a me-diator for controlled radical polymeri-zation of vinyl acetate (VAc) and com-pared with the analogue [CoII ACHTUNGTRENNUNG(acac)2](1; acac= acetylacetonato). A relativelywell controlled process occurs, after aninduction time, with 2,2’-azobis(4-me-thoxyl-2,4-dimethylvaleronitrile) (V-70)as radical initiator at 30 8C. However,whereas the polymerization essentiallystops after about six initiator half-livesin the presence of 1, it continues with afirst-order rate law in the presence of4. The successful simulation of the ki-netic data shows that 4 operates simul-taneously by associative (degenerativetransfer, DT) and dissociative (organo-

metallic radical polymerization,OMRP) mechanisms. The occurrenceof OMRP was confirmed by an inde-pendent polymerization experimentstarting from an isolated and purified[CoACHTUNGTRENNUNG(tmhd)2] ACHTUNGTRENNUNG(PVAc) macroinitiator.The polymer molecular weight evolveslinearly with conversion in accordancewith the expected values for one chainper Co atom when DT is the predomi-nant mechanism and also during thepure OMRP process; however, obser-vation of stagnating molecular weights

at long reaction times with concomitantbreakdown of the first-order rate lawfor monomer consumption indicates acompetitive chain-transfer process cat-alyzed by an increasing amount of CoII.In the presence of external donors L(water, pyridine, triethylamine) the DTpathway is blocked and the OMRPpathway is accelerated, and polymeri-zation with complex 4 is then aboutfive times slower than with complex 1.The reversal of relative effectiveOMRP rate constants keff (4>1 in theabsence of external donors, 4<1 intheir presence) is rationalized throughcompetitive steric effects on CoIII�Cand CoII�L bond strengths. Thesepropositions are supported by 1H NMRstudies and by DFT calculations.

Keywords: cobalt · density func-tional calculations · polymerization ·radical reactions · reaction mecha-nisms

[a] Dr. Y. Champouret, Dr. J.-C. Daran, Prof. R. PoliCNRS; LCC (Laboratoire de Chimie de Coordination)Universit� de Toulouse; UPS, INPT205, route de Narbonne, 31077 Toulouse (France)Fax: (+33) 561553003E-mail : rinaldo.poli@lcc-toulouse.fr

[b] Dr. K. S. S. Kumar, Dr. Y. GnanouLaboratoire de Chimie des Polym�res OrganiquesENSCPB-CNRS-Universit� Bordeaux, 116, Avenue Pey Berland, 33607 Pessac Cedex (France)Fax: (+33) 561553003E-mail : rinaldo.poli@lcc-toulouse.fr

[c] Prof. R. PoliInstitut Universitaire de France103 bd Saint-Michel, 75005 Paris (France)

Supporting information for this article is available on the WWWunder http://dx.doi.org/10.1002/chem.200802388.

� 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2009, 15, 4874 – 48854874

complexes, pioneered by J�r�me and co-workers with [Co-ACHTUNGTRENNUNG(acac)2] (1; acac =acetylacetonato),[13–18] has been particular-ly successful. Diblock copolymers of VAc with N-vinylpyrro-lidone,[19] styrene,[20] and acrylonitrile[21,22] have also been ob-tained by using [Co ACHTUNGTRENNUNG(acac)2]. The fluorinated analogues [Co-ACHTUNGTRENNUNG(acac-F3)] (2 ; acac-F3 =1,1,1-trifluoropenta-2,4-dionato) and[Co ACHTUNGTRENNUNG(acac-F6)2] (3 ; acac-F6 = 1,1,1,5,5,5-hexafluoropenta-2,4-dionato) were investigated by Matyjaszewski et al. for VAcand related monomers.[23,24] Whereas compound 2 exerts alevel of control comparable to that of 1, compound 3 yieldshigher molecular weights and higher polydispersities.

Recently, we have shown that the control exerted by [Co-ACHTUNGTRENNUNG(acac)2] on VAc polymerization occurs by degenerativetransfer (DT) when no additional Lewis bases are present,[25]

but it switches to reversible deactivation (RD) and morespecifically to a process that we have termed “organometal-lic radical polymerization” (OMRP)[26] in the presence ofdonor ligands such as pyridine, NEt3, or water (seeScheme 1).[25,27] Porphyrin-based cobalt systems have alsobeen shown to function either by OMRP[28–31] or by degener-ative transfer,[32–36] depending on conditions and on the

nature of the monomer. Furthermore, the two mechanismshave been shown to coexist for the polymerization of acrylicacid in water.[35]

It has been argued that OMRP should be particularlysuitable for controlling difficult monomers,[26] because thesteric effect of the ligand coordination sphere can be adjust-ed with a view to fine-tuning the metal–carbon bondstrength in a range suitable for control of any desired mono-mer. The additional advantage of metal complexes is the va-riety of available transition metals and ligands, which can betailored to mediate polymerization of the monomer ofchoice. We have already proven the validity of this principleby using a family of half-sandwich chromium b-diketimi-nates [CpCr ACHTUNGTRENNUNG{ArNC(Me)CHC(Me)NAr}], for which thesteric bulk of the aryl substituent (Ar=2,6-C6H3Me2, 2,6-C6H3iPr2) was shown to significantly affect the strength ofthe CrIII�R bond in the OMRP dormant chain.[37]

Here we report the use of [Co ACHTUNGTRENNUNG(tmhd)2] (4 ; tmhd =2,2,6,6-tetramethylhepta-3,5-dionato) in the radical polymerizationof vinyl acetate, carried out in the presence of 2,2’-azobis(4-methoxyl-2,4dimethylvaleronitrile) (V-70) as radical sourceat 30 8C. Introduction of the bulkier tBu group allows thecomplex to bring about the OMRP mechanism, even in theabsence of axial bases, although the process can be furtheraccelerated by axial bases. The studies carried out in the ab-sence of axial base have also revealed coexistence of DTand OMRP, as well as weak activity of the CoII complex as achain-transfer catalyst, which becomes significant when itsconcentration increases.

Results and DiscussionBulk polymerization of vinyl acetate in the absence of exter-nal base : Polymerization of vinyl acetate (VAc) initiated byV-70 in the presence of complex 4 was initially investigatedin neat monomer at different V-70/Co ratios. The conversionas a function of time is shown graphically in Figure 1 (seethe Supporting Information, Table S1). All polymerizationsare characterized by an induction time. For the experimentsrun with a larger excess of primary radical (R0) source (i.e. ,when V-70/Co�1), this corresponds to the time necessary

Scheme 1.

Figure 1. Conversion as a function of time for bulk VAc polymerizationinitiated by V-70 at 30 8C. The different data refer to different V-70/Coratios (values given next to each curve).

Chem. Eur. J. 2009, 15, 4874 – 4885 � 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 4875

FULL PAPER

for the primary radicals, after insertion of one or more mol-ecules of monomer, to convert all [Co ACHTUNGTRENNUNG(tmhd)2] to theorganocobalt ACHTUNGTRENNUNG(III) product [(tmhd)2CoIII ACHTUNGTRENNUNG(VAc)nR0]. In agree-ment with this view, the induction time is shorter for higherV-70/Co ratios. This process is identical to that exhibited by[Co ACHTUNGTRENNUNG(acac)2] (Scheme 1). Using the literature t1/2 value for V-70 decomposition at 30 8C[38] and the stoichiometry of oneradical chain per cobalt atom, the observed inductions timeslead to a calculated initiator efficiency f in the range 0.35–0.55. In agreement with the occurrence of a DT mechanism,the effective polymerization rate constant measured imme-diately after the induction period increases with increasinginitiator concentration.

Given that the efficiency factor is not greater than about0.5, the results of the experiments run with 0.7 and especial-ly 0.6 equiv of V-70 per Co atom cannot be rationalized onthe basis of a DT mechanism. Indeed, an insufficientamount of primary radicals is produced to convert all CoII

to CoIII in these experiments. Note also that these two ex-periments involve sustained polymer growth after 60 h fromthe beginning of the polymerization, which corresponds toabout six half-lives of the radical initiator. As is confirmedby experiments described below, no significant amount ofnew primary radicals is generated beyond 60 h at 30 8C.Therefore, these experiments suggest that polymerizationcan also occur by reversible activation/deactivation of theCoIII-capped dormant chains. The level of control is not asgood as that reported for [Co ACHTUNGTRENNUNG(acac)2],[13] but the Mn valuesshow the expected linear growth with respect to conversion,though always greater than the expected values for onechain per cobalt atom. In addition, the molecular weight dis-tributions are relatively narrow (see Figure 2).

Polymerization of vinyl acetate in toluene solution : As men-tioned in the Introduction, increasing the steric bulk of theligand coordination sphere should decrease the metal–carbon bond strength, and the above experiments with a lowV-70/Co ratio appear consistent with polymerization byOMRP. It was previously established[25] that [Co ACHTUNGTRENNUNG(acac)2]-mediated VAc polymerization switches from DT to OMRP

in the presence of axial bases, but the action of compound 1via the OMRP mechanism in the absence of external baseswas not proven. The reported[13] ability of an isolated andpurified [CoACHTUNGTRENNUNG(acac)2]-capped chain (macroinitiator) to under-go chain extension in the absence of new radical source ap-peared to result from water coordination, caused by han-dling of the macroinitiator in moist air. Indeed, when theisolation and purification procedures were performed underan inert atmosphere, the subsequent chain extension processin the absence of new radicals was very slow, while the rateincreased on addition of either water (facilitating theOMRP mechanism) or a radical source (triggering the DTmechanism).[27]

To confirm the existence of OMRP in the [Co ACHTUNGTRENNUNG(tmhd)2]-mediated polymerization and also, for comparison purposes,whether any OMRP may occur in the [CoACHTUNGTRENNUNG(acac)2]-mediatedpolymerization, we took a different approach, consisting ofslowing down the DT mechanism in such a way that incom-plete conversion is reached in the time needed to fully ex-haust all radical initiator. This can be easily accomplishedby diluting the reaction mixture. Then, if the polymerizationcontinues beyond this point, the existence of an OMRPmechanism is proven.

Figure 3 shows the evolution with time of the conversionobtained for 50:50 mixtures of monomer and toluene in thepresence of two equivalents of V-70 relative to cobalt (seethe Supporting Information, Table S2). As can be clearlyseen in the enlargement of the initial stages of the polymeri-zation process, the conversion evolves essentially along thesame curve during the first 60 h (ca. 6 half-lives of the V-70initiator), after the initial induction time and through theregion of polymer growth by DT. This is expected, becausethe polymerization rate for the degenerative transfer mecha-

Figure 2. Dependence of the Mn and Mw/Mn of PVAc on monomer con-version for VAc polymerization at 30 8C. The straight line represents thetheoretical Mn. The meaning of the different symbols is the same as inFigure 1.

Figure 3. Time dependence of ln([M]0/[M]) for polymerization of VAc at30 8C in toluene solution (50 % v/v). VAc/V-70/Co =500/2/1. Squares:polymerization mediated by 4 ; diamonds: polymerization mediated by 1.

www.chemeurj.org � 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2009, 15, 4874 – 48854876

R. Poli, Y. Gnanou et al.

nism is the same as that of free-radical polymerization andshould not depend on the nature or concentration of thetransfer agent. However, the polymerization essentiallystops after about 60 h for the process mediated by 1 (ca. 3 %additional conversion in the subsequent 265 h), whereas itcontinues at a reduced rate for the process mediated by 4,which clearly suggests that an OMRP mechanism remainsoperational and that this mechanism is much more favoredfor 4. This experiment also sets a new upper limit for therate of OMRP of VAc mediated by compound 1 under an-hydrous conditions (keff = 1.374 � 10�4 h�1 in 50 % toluene).

Figure 4 compares the VAc conversion under the sameconditions at different dilutions with toluene (data obtainedin toluene/VAc= 25/75 are collected in Table S3, Supporting

Information). The induction time is longer in the experimentcarried out at higher dilution. This is caused by the expectedreduced initiator efficiency f at higher dilution. From thesimulations (see below) the efficiency factor was calculatedas 0.365, 0.325, and 0.285 for toluene/VAc=0/100, 25/75,and 50/50, respectively. The polymerization rate after the in-duction period is greater for the more concentrated mix-tures, as a result of greater [VAc] and f. Consequently, whenthe transition from DT to OMRP occurs (ca. 60 h), the con-version is greater for the more concentrated mixture.

A more detailed data analysis was carried out as follows(this is summarized here only for the experiment with tolu-ene/VAc=50/50; for additional details and for the othersimulations, see the Supporting Information). Figure 5shows that between 60 h (the time at which essentially all V-70 is consumed) and 164 h, a period of time during whichthe conversion increases approximately from 30 to 60 %, themonomer consumption follows a first-order rate law, as ex-pected for the OMRP mechanism. The deviation observedat greater times is attributed to breakdown of the persistentradical effect (PRE), caused by the extremely long reactiontimes. From the slope of this line, we derive an effective rateconstant for the OMRP process (keff = 5.09 � 10�3 h�1, over30 times the upper estimate for the OMRP effective rate in

the presence of [Co ACHTUNGTRENNUNG(acac)2]). Note that the breakdown ofthe PRE implies an increase in [4], indirect evidence ofwhich is provided by the onset of catalytic chain-transfer ac-tivity (vide infra).

Taking into account that OMRP may also take placeduring the induction period, though the radical activationequilibrium is repressed by the excess of CoII (the[CoIII]/ACHTUNGTRENNUNG[CoII] ratio increases from 0 at t=0 to a maximumvalue at the onset of the DT process, tind), and that the DTand OMRP mechanisms operate simultaneously after tind,equations were derived for the time evolution of ln([M]0/[M]) and fitted to the data during the entire experiment (upto 164 h). An adjustment had to be made to account for thesmall monomer consumption before the induction time (2 %at 5 h), because this could not properly be accounted for bythe intervention of OMRP. We assume that this amount re-sults from the small amount of monomer that adds to theprimary radicals before their trapping by 4 to form the ini-tial [(tmhd)2CoACHTUNGTRENNUNG(VAc)nR0] oligomers. Indeed, the corre-sponding [CoACHTUNGTRENNUNG(acac)2] system was shown to yield [(acac)2Co-ACHTUNGTRENNUNG(VAc)nR0] oligomers with n�3, which were isolated andcharacterized, under conditions of a large excess of 1 withrespect to monomer.[27] The propagation rate constant kp

was fixed at the literature value (117 s�1m�1),[38] whereas the

rate constant for initiator decomposition kd, efficiency factorf, and termination rate constant kt were freely optimized.The curve fitting to the data gave optimized values for theserate constants (kd =0.072 h�1, kt =1.74 � 108 s�1

m�1) in close

agreement with the literature,[38] attesting to the self-consis-tency of the mechanistic model. The simulated conversionversus time curves are shown in Figure 4.

A final test for the presence of OMRP was carried out asfollows. A new experiment was carried out exactly as shownin Figure 3 (50/50 toluene/VAc mixture), but the polymeri-zation was stopped after 41 h, corresponding to a conversionof 13 % (in good agreement with the trend shown inFigure 3). The resulting polymer was isolated by precipita-tion and washed to eliminate all residual V-70. It was thenadded to the same amount of fresh VAc/toluene (1/1) andthe mixture warmed again to 30 8C. The results are collectedin Table 1. The successful resumption of polymerizationdemonstrates that the isolated polymer is indeed a macroini-

Figure 4. Comparison of VAc polymerization mediated by 4 at differentdilutions. VAc/V-70/Co =500/2/1; T=30 8C. VAc/toluene (v/v) =100/0 (di-amonds); 75/25 (triangles); 50/50 (squares). The solid lines are the simu-lated curves (see text).

Figure 5. First-order plot of the VAc polymerization mediated by 4. VAc/V-70/Co =500/2/1; T= 30 8C; toluene/VAc (v/v)=50/50.

Chem. Eur. J. 2009, 15, 4874 – 4885 � 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 4877

FULL PAPERCoexistence of Three Polymerization Mechanisms

tiator and that a reversible termination mechanism (OMRP)must be operating. Comparison of the first-order kinetics inFigure 6 yields approximately the same effective rate con-stant, that is, essentially all cobalt-containing chain endshave retained their activity through the polymer isolationand purification procedure.

With regard to the average molecular weights and molec-ular weight distribution (data in Supporting Information,Figure 2, and Table 1), it is notable that Mn,SEC grows linear-ly with conversion (and is only slightly greater than Mn,theo)up to the point in the OMRP region where the monomerconsumption follows first-order kinetics, but falls below thetheoretical value at longer times, when PRE breaks down.This is clearly shown in the progression of the GPC traces inFigure 7. The major expected phenomenon when the PREbreaks down is an increase in concentration of CoII complex4, which may also operate as a chain-transfer catalystthrough abstraction of a hydrogen atom and formation of anintermediate CoIIIH complex. CoII complexes are wellknown to serve as catalysts for catalytic chain transfer(CCT).[39–41]

As long as the DT mechanism dominates, the excess freeradicals that are continuously injected into solution by thedecomposition of V-70 represses the equilibrium dissocia-tion of the organocobalt ACHTUNGTRENNUNG(III) dormant chains, and thus theconcentration of CoII complex (the CCT agent) is kept low.The role of DT in the suppression of CCT was also pointed

out by Wayland et al. for porphyrin CoII systems.[34] Whenno new radicals are injected into solution, the CoII concen-tration increases and CCT starts to compete with a con-trolled chain growth, but only a rather low CoII concentra-tion is enforced by the unfavorable dissociation equilibrium;this results from the strong CoIII�PVAc bond, as shown bythe rather slow polymerization rate. However, the CoII con-centration continuously increases as a result of irreversibleterminations and becomes substantial at very long reactiontimes. Since the incidence of this phenomenon remainsscarce and the phenomenon becomes effective only whenthe polymer chains are already rather long, direct evidencefor the formation of unsaturated chain ends by NMR spec-troscopy or MALDI-TOF MS could not be obtained. Notethat CCT may occur at a significant rate at the beginning ofthe induction period, when [CoII]/ ACHTUNGTRENNUNG[CoIII] is high. However,NMR experiments specifically conceived to probe the for-mation of small oligomers during the induction period didnot provide any positive evidence.

In conclusion, three different functions have been high-lighted for complex 4 in VAc polymerization: 1) reversibletrapping of the growing PVAc radical chain (OMRP mecha-nism); 2) associative exchange of the growing PVAc radicalchains (DT mechanism); 3) H-atom transfer from the PVAcradical chains to yield catalyzed chain transfer (CCT). Pro-cesses 1 and 2 may coexist during the induction period andat long reaction times, whereas all processes can coexistafter tind and as long as new radicals are being injected intosolution from the initiator, but their relative importancechanges with time. Process 3 is minimized by the excess offree radicals and its relative importance increases with time,whereas process 2 decreases exponentially, together with therate of generation of new radicals. The greater incidence ofOMRP for VOAc polymerization mediated by 4, relative tothat mediated by 1, is consistent with a weaker Co�PVAcbond in the presence of the bulkier tmhd ligand. This propo-sition has been verified by DFT calculations (see below).Note that the coexistence of associative (degenerative trans-fer) and dissociative (reversible deactivation) mechanismshas previously been highlighted not only for the cobalt-mediated radical polymerization of acrylic acid by a cobalt

Table 1. Polymerization of VAc in toluene solution (50 % v/v) initiatedby a PVAc-[Co ACHTUNGTRENNUNG(tmhd)2] macroinitiator.[a]

Time [h] Conversion [%] 10�4 Mn,SEC 10�4 Mn,theo Mw/Mn

41 13.0 0.55 0.56 1.1465 19.2 1.00 0.83 1.2090 28.8 1.51 1.24 1.17136.75 40.3 1.76 1.73 1.21161 45.9 1.87 1.98 1.25233 52.3 2.03 2.25 1.19

[a] Conditions: VAc/V-70/4=500/2/1; T=30 8C; polymerization stoppedafter 41 h (13 % conversion), polymer precipitated and washed, thenpolymerization resumed by addition of monomer and toluene (1/1 v/v).The times shown include the time for macroinitiator preparation (41 h).

Figure 6. Comparative kinetics in the OMRP region for the VAc poly-merizations initiated by V-70/4 (from Figure 5, squares) and by thePVAc-[Co ACHTUNGTRENNUNG(tmhd)2] macroinitiator (Table 1, circles).

Figure 7. SEC chromatograms for PVAc from polymerization controlledby 4. VAc/V-70/Co =500/2/1; T =30 8C; toluene/VAc =50/50 (v/v).

www.chemeurj.org � 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2009, 15, 4874 – 48854878

R. Poli, Y. Gnanou et al.

porphyrin system,[35] but also for organotellurium-mediatedliving radical polymerization (TERP) of a variety of vinylmonomers.[42]

Polymerization of vinyl acetate in the presence of an exter-nal base : Additional experiments on [CoACHTUNGTRENNUNG(tmhd)2]-mediatedVAc polymerization were carried out in the presence of anexternal base (pyridine, triethylamine, or water), in order toevaluate the ability of the coordinating ligand to shut downthe DT mechanism and to speed up the OMRP mechanism,by analogy with the previously reported polymerization con-trolled by compound 1.[25] For the purpose of direct compari-son, polymerizations were conducted under conditions iden-tical to those previously used for 1 (bulk, 30 8C, VAc/V-70/Co/L=500/0.7/1/30). The results are collected in Table 2. In

brief, the behavior is qualitatively identical to that previous-ly observed for polymerization controlled by 1 in terms ofthe time evolution of the conversion (see Figure 8): 1) thereis no induction time in the py and NEt3 experiments (for theH2O experiment, vide infra); 2) polymerization obeys first-order kinetics; 3) the effective polymerization rate decreasesas the added ligand L varies in the order H2O>py> NEt3>

no ligand (before the induction time); 4) there is no increasein polymerization rate after all the CoII has been trans-formed into CoIII (i.e., when excess radicals relative to Costart to be injected into solution). The last observationproves that L addition completely shuts down the DT mech-anism.

To explain the short induction time and nonlinearity ofthe first-order plot observed for the H2O experiment (seeFigure 8), we advance the hypothesis that 4 does not reactrapidly with H2O, because it is confined in the monomerphase, with which water is immiscible. Thus, polymerizationstarts only after a sufficient amount of 4 has reacted withH2O, and the polymerization rate increases as more adductis generated.

The Mn values are considerably greater than theory, be-cause the polymer samples are withdrawn when only a frac-tion of the CoII complex has been transformed into CoIII,but the average values increase with conversion, and thepolydispersities are relatively narrow, similar to those of thepolymer obtained in the absence of added ligand, which at-tests to the similar controlling ability of PVAc�[Co-ACHTUNGTRENNUNG(tmhd)2](L) and PVAc�[CoACHTUNGTRENNUNG(tmhd)2]. Thus, the major effectof L is to speed up the OMRP process, as already estab-lished for polymerization controlled by 1.[25,27]

It is of interest to compare the effective polymerizationrate constants in the presence of added ligand for com-pounds 1 and 4. The approximate pseudo-first-order rateconstants derived from the data in Figure 8 are 5.50 �10�3 h�1 for L=py and 6.67 � 10�4 h�1 for L=NEt3. Thesevalues are about a factor of five smaller than those of poly-merization mediated by 1 under identical conditions (2.25 �10�2 h�1 for L= py and 1.40 � 10�3 h�1 for L=NEt3).[25] Thisbehavior is opposite to what is observed in the absence ofadded donor: compound 1 gives faster polymerization thancompound 4 in the presence of coordinating ligands, butslower polymerization in the absence of ligands. The reasonfor this unexpected behavior is that the weaker Co�PVAcbond in the order tmhd<acac is more than compensated bya the weaker Co�L bonds in the same order for the Lewisbase stabilized [Co{RC(O)CHC(O)R}2(L)2] (R=Me, tBu)species. The strength of this interaction was investigated inmore detail by experimental and computational techniquesfor the specific case of L=py (see below).

Experimental study on the [Co ACHTUNGTRENNUNG(tmhd)2]–pyridine interac-tion : trans-[Co ACHTUNGTRENNUNG(tmhd)2(py)2] was obtained in the form ofsingle crystals on interaction between 4 and a large excessof pyridine. A view of the molecular geometry, obtained byX-ray diffraction, is shown in Figure 9, and selected structur-al parameters are listed in Table 3. The compound crystalli-

Table 2. Polymerization data for bulk VAc polymerization mediated by 4in the presence of added ligands (L).[a]

L Time [h] Conversion [%] 10�4 Mn,SEC 10�4 Mn,theo Mw/Mn

H2O 4 1.3 – – –H2O 7 3.5 – – –H2O 19 50.4 6.49 2.17 1.31H2O 24 79.0 9.86 3.40 1.68py 4 3.5 – – –py 7 4.7 – – –py 20 13.7 2.33 0.56 1.25py 44 25.6 2.46 1.10 1.60py 68 32.0 3.44 1.38 1.39py 92 37.0 3.50 1.60 1.45NEt3 4 2.1 – – –NEt3 19 2.5 – – –NEt3 51 4.2 – – –NEt3 67 4.5 – – –NEt3 91 5.0 – – –

[a] Conditions: VAc/V-70/Co/L= 500/0.7/1/30; T= 30 8C.

Figure 8. Bulk VAc polymerization in the presence of different added li-gands. Conditions are as shown in Table 2. The data obtained in the ab-sence of added base under the same conditions (from Figure 1) are alsoshown for comparison.

Chem. Eur. J. 2009, 15, 4874 – 4885 � 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 4879

FULL PAPERCoexistence of Three Polymerization Mechanisms

zes with the molecule sitting on an inversion center, andthus all trans angles are linear by symmetry. The ground-state structure does not reveal any particular steric encum-brance, since the Co�O and Co�N distances are very similarto those of the previously described trans-[Co ACHTUNGTRENNUNG(acac)2(py)2]analogue.[43] In fact, the Co�N bond length in the tmhd com-plex is even slightly shorter than that reported for the acacanalogue.

Further information on the 4–py interaction was obtainedby 1H NMR spectroscopy. The 1H NMR spectra of solutionsof 4 containing variable amounts of pyridine are shown inFigure 10. The spectra are rather complex, but the changesin position and relative intensity of the various resonancesas a function of py/4 ratio helps to make a few important as-signments. Peak a at d�12.0 is already present in the spec-trum of the starting compound. It remains observable at thesame chemical shift after addition of pyridine and its intensi-ty decreases until it disappears at a py/4 ratio greater than 3.Concomitantly, a new peak at d�20.0 (labeled b) grows inand its relative intensity no longer varies after reaching apy/4 ratio of about three. The position of this peak is also in-variable. There are no other peaks the intensity of which re-mains invariant with the py/4 ratio (other than the solventpeak and some diamagnetic impurities in the region be-tween d= 0 and 1, removed from Figure 10 for clarity). It is

therefore relatively straightforward to assign the resonanceat d=12.0 to the tBu protons of 4 and that at d= 20.0 to thesame protons of the bis-pyridine adduct, which is the isolat-ed product (vide supra). The methyne protons of the tmhdligand remains unobserved, certainly because of a greaterFermi contact shift and broadening, and a smaller intensity.The disappearance of 4 only after the addition of at least3 equiv of py clearly indicates that the py-addition reactionis equilibrated [Eq. (1)]. On the other hand, the independ-ent observation of the resonances of reagent and productand their chemical shift invariance with respect to the py/4ratio indicates that this association/dissociation process ismuch slower than the chemical shift difference (k !8 �250 Hz=2000 s�1).

½CoðtmhdÞ2� þ 2 pyÐ trans-½CoðtmhdÞ2ðpyÞ2� ð1Þ

For three resonances the chemical shift changes with thepy/4 ratio, and at the same time their intensity increaseswith increasing py/4 ratio. They must therefore be associatedwith free pyridine (ortho, meta, and para protons). At largepy/4 ratio, they converge toward the typical chemical shiftsof the free py protons. Their assignment is based on theirrelative 2:2:1 intensity (p resonance) and extent of Fermicontact shift (o resonance). Note that these peaks are al-ready observable for a py/4 ratio of only 0.5, in agreementonce again with the equilibrium of Equation (1). The strongdependence of the chemical shift on py/4 ratio shows thatfree py exchanges continuously with the paramagnetic CoII

complex. This process must also be at the slow exchangelimit, because the difference Dn between the resonances of

Figure 9. ORTEP of the molecular geometry of trans-[Co ACHTUNGTRENNUNG(tmhd)2(py)2].Ellipsoids are drawn at 30 % probability.

Table 3. Selected distances [�] and angles [8] for trans-[Co ACHTUNGTRENNUNG(acac)2]·2pyand trans-[Co ACHTUNGTRENNUNG(tmhd)2]·2 py.[a]

trans-[Co ACHTUNGTRENNUNG(acac)2(py)2] trans-[Co ACHTUNGTRENNUNG(tmhd)2(py)2]X-ray[b] DFT X-ray DFT

Co�O 2.034(3) 2.050 2.036(12) 2.044Co�N 2.186(11) 2.205 2.1626(15) 2.211O�C 1.241(6) 1.269 1.264(2) 1.271O-Co-O (cis) 89.8(2) 90.21 87.76(5) 89.29O-Co-O (trans) 179.2(2) 177.84 180 178.62N-Co-N 180 180.00 180 179.88

[a] Only the average values are given for chemically equivalent parame-ters. [b] From ref. [43].

Figure 10. 1H NMR spectra of C6D6 solutions containing 4 and pyridineat variable py/4 ratio.

www.chemeurj.org � 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2009, 15, 4874 – 48854880

R. Poli, Y. Gnanou et al.

coordinated and free pyridine protons is certainly greaterthan the 8 ppm difference between the tBu protons in 4 andin the bis-pyridine adduct. The resonances of the coordinat-ed pyridine protons in complex trans-[Co ACHTUNGTRENNUNG(tmhd)2(py)2]]could not be detected in the 1H NMR spectrum. The spindensity can be readily transmitted to these protons via aro-matic-ring conjugation, whereas the tBu groups of the tmhdligand are separated from the conjugated b-diketonatoCoO2C3 ring by three saturated bonds. Thus, the electronspin density on the coordinated py protons has a stark effecton the diamagnetic resonances of free py, even though theyare far from the coalescence region.

Other peaks observable in the 1H NMR spectrum couldnot be clearly attributed, for instance, a paramagneticallyshifted peak observed at d=11 for a py/4 ratio of 0.4, whichmoves upfield and converges toward d�4 for large py/4ratios. The relative intensity of this peak seems constant,like that of complex trans-[Co ACHTUNGTRENNUNG(tmhd)2(py)2], but contrary tothe tBu resonance its position changes with the py/4 ratio(see Figure 10). Other small peaks are also present, also inthe upfield region between �5and �10 ppm, but their relativeintensity and changes in chemi-cal shift with py/4 ratio did notallow us to make reasonablehypotheses about their origin.We cannot exclude that someof these peaks are generated byimpurities (although the solu-tions were prepared from abatch of clean single crystals),since the compound is quitesensitive in solution.

The most interesting conclu-sion from this study comesfrom a comparison with thepreviously described 1H NMRstudy on the 1/py interaction.[25]

In that case, the three resonan-ces assigned to the free py pro-tons are much more affected bythe electron spin density com-pared with the resonances in the 4/py mixture at the samepy/Co ratio. For instance, for a ratio of 3, the resonances areobserved at d�33 (o), 8 (m), and �2 (p) for 1-py,[25] whereasthe same resonances are much closer to the diamagneticregion (d= 13.7, 7.4 and 5.3, respectively) for 4-py (seeFigure 10). This clearly indicates that equilibrium (1) ismuch less shifted toward the formation of the bis-pyridinecomplex for compound 4 than for compound 1. In otherwords, the formation of the bis-adduct is thermodynamicallyless favorable for 4 than for 1.

DFT calculations : Since it is important to evaluate how thesteric effect of the tBu versus Me groups on the b-diketo-nate ligands is reflected in the CoIII�PVAc and CoII�L bondstrengths, calculations were conducted at the full quantum

mechanical level on the full molecule, without any simplifi-cation for the b-diketonate and L ligands. The polymerchain, on the other hand, was modeled by a CH-ACHTUNGTRENNUNG(OOCCH3)CH3 group. The calculations involving an exter-nal Lewis base were restricted to the pyridine system.

In a previous study, we compared two different hybridfunctionals, B3LYP and B3PW91*, and concluded that thelatter provides quantitatively more reliable results.[44] There-fore, calculations for the present study were only carried outwith B3PW91*. The contribution of exact exchange is re-duced in this functional, which therefore provides better es-timates of bond energetics for processes involving first-rowtransition metals in which a spin-state change is in-volved.[45,46] A few bis-acetylacetonato systems have alreadybeen reported in previous studies, but with the energeticsgiven only at the B3LYP level[27] or at the B3PW91* levelon the fixed B3LYP-optimized geometries.[44] For better con-sistency, all geometries were fully reoptimized here. Previ-ous computational work also established the preferred ste-reochemistry of all systems (see Scheme 2),[25] so no further

computational exploration of this issue was necessary in thepresent case. The chelating nature of the ultimate monomerunit in the CoIII-capped dormant chain has been establishedby combined experimental and computational study for therelated bis-acac system, which included the 1H NMR investi-gation of an isolated short oligomer, namely, (acac)2Co�ACHTUNGTRENNUNG(VAc)nR0 with n�3.[27] The tetrahedral geometry of [Co-ACHTUNGTRENNUNG(tmhd)2] (4) is known from a previous X-ray diffractionstudy,[47] while the trans-octahedral geometry for [Co-ACHTUNGTRENNUNG(tmhd)2(py)2] was verified in the present investigation (videsupra).

The energetic results are graphically shown in Figure 11and the Cartesian coordinates and selected bond lengthsand angles of the optimized geometries are provided in theSupporting Information. Selected bonding parameters for

Scheme 2.

Chem. Eur. J. 2009, 15, 4874 – 4885 � 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 4881

FULL PAPERCoexistence of Three Polymerization Mechanisms

trans-[Co ACHTUNGTRENNUNG(acac)2(py)2]] and trans-[Co ACHTUNGTRENNUNG(tmhd)2(py)2]] are alsoreported in Table 3 and compared with those established byexperiment. The agreement supports the suitability of thecomputational level. The steric effect of the tBu group is notsignificantly reflected in the bond lengths (Co�OACHTUNGTRENNUNG(acac),Co�C, and Co�OACHTUNGTRENNUNG(ester) distances change only marginallyon going from the acac to the tmhd system). On the otherhand, breaking the CoIII�C bond costs less in enthalpy forthe tmhd derivative (15.8 kcal mol�1 for tmhd vs. 17.2 kcalmol�1 for acac, DH =1.4 kcal mol�1). Part of this enthalpydifference is related to easier opening of the five-memberedchelate ring, and the rest comes from homolytic rupture ofthe CoIII�C bond in the five-coordinate complex (seeFigure 11).

In recently published calculations on complexes [CpCr-ACHTUNGTRENNUNG{ArNC(Me)CHC(Me)NAr}{CH ACHTUNGTRENNUNG(OCOCH3)CH3}],[37] a muchgreater bond weakening for the CrIII�C bond (by8.7 kcal mol�1 as Ar is changed from Ph to 2,6-C6H3Me2) isaccompanied by a significant bond lengthening of 0.015 �.In the present case, the increase in steric bulk is not reflect-ed in lengthening of the metal–ligand bond. Therefore, thebond weakening must be related to a more subtle effect ofincreased ligand–ligand repulsions, reflected in the adoptionof less favorable conformations, rather than longer metal–ligand bonds. At any rate, the computational result agreeswith the experimental observation of a faster radical activa-tion from the dormant chain in the order tmhd>acac, in theabsence of external donors. Recent calculations on a bis(pyr-idylimino)isoindolato system with R= CHEtCO2Me re-

vealed no significant effect of the pyridyl and indolato sub-stituents on the CoIII�R bond-dissociation energy.[48]

In the presence of external ligands, the dormant chainmay open the five-membered chelate to yield a six-coordi-nate adduct, depending on the donor strength of the ligand.For pyridine, the calculations indicate that this adduct isslightly enthalpically favored with respect to the chelatecomplex (by 1.2 and 1.0 kcal mol�1 for acac and tmhd, re-spectively; see Figure 11). The dormant chain, however, mayremain as a ligand-free chelate complex, because of the un-favorable entropy term associated with addition of pyridineto the system. This term cannot be evaluated quantitativelyin the condensed phase. Note that although the chelateopening step is easier for the tmhd complex, subsequent pycoordination stabilizes the acac system to a greater extent.In this case, a steric effect on the strength of the Co�pybond is supported by the observation of a significantlylonger Co�N bond (by 0.016 �) for the tmhd derivative.

In the presence of pyridine, the CoII complex will exist insolution in the form of the bis-pyridine adduct, as proved byits isolation in crystalline form and by the NMR study. Thecalculations show that pyridine coordination entails a signifi-cant enthalpic stabilization, in agreement with the accelerat-ing effect on the OMRP mechanism. According to the cal-culations, the bis-pyridine adducts are essentially at thesame relative enthalpy with respect to the dormant chain forthe acac and tmhd systems. This means that the more diffi-cult CoIII�C bond breaking for the acac system is compen-sated by a more favorable CoII�py bond formation havingthe same net enthalpy effect. The weaker CoII�py bond forthe tmhd system may once again be associated with a stericeffect: the Co�N bond is 0.006 � longer in the bulkier tmhdsystem.

The computed relative enthalpy change does not perfectlyreproduce the faster OMRP in the order acac> tmhd in thepresence of pyridine. However, the ratio of effective rateconstants (4.09 in favor of 1 at 303 K) corresponds to a free-energy difference of only 0.85 kcal mol�1, a small value incomparison with the computational accuracy. The discrepan-cy may be associated with the neglect of entropic effectsand solvation effects, which may be subtle, as well as to theaccuracy of the chosen computational level (functional andbasis sets). The calculation of bond strengths for weakbonds is also significantly affected by the basis set superpo-sition error (BSSE),[49] which should, however, be largelycompensated in the calculation of the enthalpy differences.The most valuable contribution of this computational studyis not in the perfect reproduction of the experimentally ob-served trends, but rather lies in a better insight into thecauses of the relative OMRP rate reversal for systems 1 and4 in the presence and absence of external donor ligands.This is unambiguously attributed to a significantly weakerCoIII�C bond for the tmhd system, which opens the OMRPpathway for VAc polymerization controlled by complex 4 inthe absence of external donors, whereas the presence of ex-ternal donors provides a greater stabilization to the CoII

system in the presence of the less-encumbering acac ligand

Figure 11. Relative enthalpy changes for processes involving compounds1 or 4, the CH ACHTUNGTRENNUNG(OOCCH3)CH3 radical, and pyridine.

www.chemeurj.org � 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2009, 15, 4874 – 48854882

R. Poli, Y. Gnanou et al.

and overrules the negative effect of the CoIII�C bondstrength.

Conclusions

We have shown that increasing the steric bulk of the ligandcoordination sphere has important effects on the competi-tive rates of different processes in metal-mediated radicalpolymerization, using the specific example of (b-diketona-to)cobalt(II) complexes, [{RC(O)CHC(O)R}2Co]. On goingfrom the acetylacetonato (R=Me) to the 2,2,6,6-tetrame-thylhepta-3,5-dionato (R= tBu) compound, the CoIII�Cbond of the metal-capped dormant chain is slightly weak-ened. This weakening effect (ca. 1.5 kcal mol�1 according toDFT calculations) allows the tBu system to contribute to thepolymer growth process with a dissociative mechanism(OMRP), whereas the Me system operates only by an asso-ciative mechanism (DT), while the OMRP component isnegligible in the absence of external Lewis bases (L). Theaddition of L molecules, however, overturns this trend be-cause the steric bulk of the b-diketonato ligands has an evengreater effect on the CoII�L bonds. Understanding how theligand coordination sphere tunes metal–carbon and metal–ligand bond strength is essential for the future developmentof systems that may be capable of controlling the polymeri-zation of more challenging monomers.

Experimental SectionMaterials : Vinyl acetate (VAc) (>99%, Alfa Aeser) was dried over cal-cium hydride, distilled under argon at 90 8C, degassed by several freeze/thaw cycles, and stored under argon. Toluene was distilled from sodiumbenzophenone ketyl. 2,2’-Azobis(4-methoxy-2,4-dimethylvaleronitrile)(V-70) was used as received. Complex [CoACHTUNGTRENNUNG(tmhd)2] (4) was synthesizedaccording to the literature procedure.[50]

Characterization : 1H NMR spectra were recorded on a Bruker ARX 250spectrometer. Deuterated benzene was degassed and stored in a Schlenktube under argon. Size exclusion chromatography (SEC) of poly(vinylacetate) was carried out in filtered THF (flow rate: 1 mL min�1) at 35 8Con a 300 � 7.5 mm PL gel 5 mm mixed-D column (Polymer Laboratories),equipped with multiangle light-scattering (miniDawn Tristar, Wyatt Tech-nology Corporation) and refractive-index (RI2000, Sopares) detectors orwith a Waters column pack (300 � 7.5 mm, Ultrastyragel 104, 103, 100 �),equipped with multiangle light scattering (miniDawn Tristar, Wyatt Tech-nology Corp.) and refractive index (Waters 410) detectors.

General procedure for the radical polymerization of vinyl acetate : Allpolymerizations were conducted by following the same experimental pro-cedure. All operations were carried out under a protective argon atmos-phere. Complex 4 (36.9 mg, 0.086 mmol) and V-70 (18.7 mg 0.061 mmol)were introduced into a Schlenk tube, followed by the addition of de-gassed vinyl acetate (4 mL, 43 mmol). The Schlenk tube was then im-mersed in an oil bath at 30 8C and aliquots were withdrawn periodicallyfor reaction monitoring by GPC. The monomer conversion was deter-mined gravimetrically after removal of the unconverted monomer underreduced pressure, and the resulting residue was used for SEC characteri-zation. For polymerizations carried out in the presence of toluene, an ad-ditional 4 mL of toluene (monomer/toluene=1/1 (v/v)) was introducedinto the Schlenk tube at the beginning of the experiment.

X-ray crystallography : Single crystals of trans-[Co ACHTUNGTRENNUNG(tmhd)2]·2py) were pre-pared by dissolving [CoACHTUNGTRENNUNG(tmhd)2] (0.200 g, 0.466 mmol) in diethyl ether(5 mL) to afford a pink solution. Addition of few drops of pyridine (ca.

10 equiv) resulted in an immediate color change to orange. The solutionwas filtered through Celite and placed at �20 8C, yielding orange crystals(96 mg, 35%). A single crystal was mounted under inert perfluoropo-lyether at the tip of a glass fiber and cooled in the cryostream of aBruker APEXII diffractometer. Data were collected with monochromat-ic MoKa radiation (l=0.71073). The structure was solved by direct meth-ods (SIR97)[51] and refined by least-squares procedures on F2 usingSHELXL-97.[52] All H atoms attached to carbon were introduced in ide-alized positions and treated as riding models in the calculations. Thestructures were drawn with ORTEP3.[53] Crystal data and refinement pa-rameters are shown in Table 4. CCDC-706747 contains the supplementa-ry crystallographic data for this paper. These data can be obtained freeof charge from The Cambridge Crystallographic Data Centre viawww.ccdc.cam.ac.uk/data_request/cif.

Computational details : All geometry optimizations were performed withthe Gaussian 03 suite of programs[54] using the B3PW91* functional. Thisis a modified version of the B3PW91 functional which combines thethree-parameter hybrid density functional method of Becke[55] with theexchange component of the Perdew and Wang 1991 functional[56, 57] andin which the c3 coefficient in the original Becke three-parameter fit tothermochemical data was changed to 0.15. The basis functions consistedof the standard 6-31G** for all light atoms (H, C, N, O), plus theLANL2DZ function, which included the Hay and Wadt effective core po-tentials (ECP)[58] for Co. The latter basis set was, however, augmentedwith an f polarization function (a= 0.8) in order to obtain a balancedbasis set and to improve the angular flexibility of the metal functions. Allgeometry optimizations were carried out without any symmetry con-straint and all final geometries were characterized as local minima of thepotential-energy surface (PES) by verifying that all second derivatives ofthe energy were positive. The unrestricted formulation was used foropen-shell molecules. The closeness of hS2i at convergence to the expect-ed value of 0.75 for the radical species and 3.75 for the spin quartet spe-cies (the greatest deviation was found for [Co ACHTUNGTRENNUNG(tmhd)2] with hS2i=3.7582)indicates minor spin contamination. All energies were corrected for zero-point vibrational energy and for thermal energy to obtain the bond disso-

Table 4. Crystal data and structure refinement for trans-[Co-ACHTUNGTRENNUNG(tmhd)2(py)2].

empirical formula C32H48CoN2O4

formula weight 583.65temperature [K] 180(2)wavelength [�] 0.71073crystal system triclinicspace group P1̄a [�]; a [8] 8.8350(4); 76.971(3)b [�] ; b [8] 9.8230(5); 73.068(2)c [�] ; g [8] 11.0047(5); 65.737(2)V [�3] 826.80(7)Z 11calcd [Mg m�3] 1.172absorption coefficient [mm�1] 0.554F ACHTUNGTRENNUNG(000) 313crystal size [mm] 0.548 � 0.213 � 0.137q range for data collection [8] 2.29–28.28reflns collected 14 132independent reflns (Rint) 4085 (0.0557)completeness to q=28.288 [%] 99.5absorption correction semiempirical from equivalentsmax./min. transmission 1.0/0.888refinement method full-matrix least-squares on F2

data/restraints/parameters 4085/0/184GOF on F2 1.079R, wR2 [I>2s(I)] 0.0410, 0.1072R, wR2 (all data) 0.0498, 0.1141largest diff. peak/hole [e ��3] 0.857 and �0.677

Chem. Eur. J. 2009, 15, 4874 – 4885 � 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 4883

FULL PAPERCoexistence of Three Polymerization Mechanisms

ciation enthalpies at 298 K. The standard approximations for estimatingthese corrections were used (ideal gas, rigid rotor, and harmonic oscilla-tor), as implemented in Gaussian 03.

Acknowledgements

Support for this work from the Agence Nationale de la Recherche (con-tract ANR No. NT05-2_42140) is gratefully acknowledged. R.P. alsothanks the Institut Universitaire de France for additional support and the“Centre Interuniversitaire de Calcul de Toulouse” (Project CALMIP) forgranting free computational time.

[1] K. Matyjaszewski, J. H. Xia, Chem. Rev. 2001, 101, 2921 – 2990.[2] M. Kamigaito, T. Ando, M. Sawamoto, Chem. Rev. 2001, 101, 3689 –

3745.[3] C. J. Hawker, A. W. Bosman, E. Harth, Chem. Rev. 2001, 101, 3661 –

3688.[4] J. Chiefari, Y. K. Chong, F. Ercole, J. Krstina, J. Jeffery, T. P. T. Le,

R. T. A. Mayadunne, G. F. Meijs, C. L. Moad, G. Moad, E. Rizzardo,S. H. Thang, Macromolecules 1998, 31, 5559 –5562.

[5] A. Dur�ault, Y. Gnanou, D. Taton, M. Destarac, F. Leising, Angew.Chem. 2003, 115, 2975 – 2978; Angew. Chem. Int. Ed. 2003, 42, 2869 –2872.

[6] V. Sciannamea, R. Jerome, C. Detrembleur, Chem. Rev. 2008, 108,1104 – 1126.

[7] T. Ando, M. Kamigaito, M. Sawamoto, Polym. Prepr. (Am. Chem.Soc. Div. Polym. Chem.) 2002, 43, 179 – 180.

[8] E. Rizzardo, J. Chiefari, R. T. A. Mayadunne, G. Moad, S. H. Thang,ACS Symp. Ser. 2000, 768, 278 – 296.

[9] E. Rizzardo, J. Chiefari, R. Mayadunne, G. Moad, S. Thang, Macro-mol. Symp. 2001, 174, 209 – 212.

[10] M. Destarac, D. Charmot, X. Franck, S. Z. Zard, Macromol. RapidCommun. 2000, 21, 1035 –1039.

[11] M. Wakioka, K. Y. Baek, T. Ando, M. Kamigaito, M. Sawamoto,Macromolecules 2002, 35, 330 – 333.

[12] M. C. Iovu, K. Matyjaszewski, Macromolecules 2003, 36, 9346 –9354.[13] A. Debuigne, J. R. Caille, R. J�r�me, Angew. Chem. 2005, 117,

1125 – 1128; Angew. Chem. Int. Ed. 2005, 44, 1101 –1104.[14] A. Debuigne, J. R. Caille, C. Detrembleur, R. Jerome, Angew.

Chem. 2005, 117, 3505 – 3508; Angew. Chem. Int. Ed. 2005, 44, 3439 –3442.

[15] A. Debuigne, J.-R. Caille, R. J�r�me, Macromolecules 2005, 38,5452 – 5458.

[16] A. Debuigne, J. R. Caille, N. Willet, R. Jerome, Macromolecules2005, 38, 9488 –9496.

[17] C. Detrembleur, A. Debuigne, R. Bryaskova, B. Charleux, R.Jerome, Macromol. Rapid Commun. 2006, 27, 37– 41.

[18] R. Bryaskova, C. Detrembleur, A. Debuigne, R. Jerome, Macromo-lecules 2006, 39, 8263 –8268.

[19] A. Debuigne, N. Willet, R. Jerome, C. Detrembleur, Macromole-cules 2007, 40, 7111 – 7118.

[20] R. Bryaskova, N. Willet, A. Debuigne, R. Jerome, C. Detrembleur,J. Polym. Sci. Part A 2007, 45, 81 –89.

[21] A. Debuigne, J. Warnant, R. Jerome, I. Voets, A. De Keizer, M. A.Stuart, C. Detrembleur, Macromolecules 2008, 41, 2353 –2360.

[22] A. Debuigne, C. Michaux, C. J�r�me, R. J�r�me, R. Poli, C. De-trembleur, Chem. Eur. J. 2008, 14, 7623 –7637.

[23] H. Kaneyoshi, K. Matyjaszewski, Macromolecules 2005, 38, 8163 –8169.

[24] H. Kaneyoshi, K. Matyjaszewski, Macromolecules 2006, 39, 2757 –2763.

[25] S. Maria, H. Kaneyoshi, K. Matyjaszewski, R. Poli, Chem. Eur. J.2007, 13, 2480 –2492.

[26] R. Poli, Angew. Chem. 2006, 118, 5180 –5192; Angew. Chem. Int. Ed.2006, 45, 5058 –5070.

[27] A. Debuigne, Y. Champouret, R. J�r�me, R. Poli, C. Detrembleur,Chem. Eur. J. 2008, 14, 4046 – 4059.

[28] B. B. Wayland, G. Poszmik, S. Mukerjee, J. Am. Chem. Soc. 1994,116, 7943 –7944.

[29] B. B. Wayland, L. Basickes, S. Mukerjee, M. Wei, M. Fryd, Macro-molecules 1997, 30, 8109 – 8112.

[30] B. B. Wayland, S. Mukerjee, G. Poszmik, D. C. Woska, L. Basickes,A. A. Gridnev, M. Fryd, S. D. Ittel, ACS Symp. Ser. 1998, 685, 305 –315.

[31] Z. Lu, M. Fryd, B. B. Wayland, Macromolecules 2004, 37, 2686 –2687.

[32] B. B. Wayland, X.-F. Fu, Z. Lu, M. Fryd, Polym. Prep. (Am. Chem.Soc. Div. Polym. Chem.) 2005, 230th ACS National Meeting, 2005,Washington.

[33] B. B. Wayland, X. Fu, C.-H. Peng, Z. Lu, M. Fryd, ACS Symp. Ser.2006, 944, 358 –371.

[34] B. B. Wayland, C.-H. Peng, X. Fu, Z. Lu, M. Fryd, Macromolecules2006, 39, 8219 –8222.

[35] C.-H. Peng, M. Fryd, B. B. Wayland, Macromolecules 2007, 40,6814 – 6819.

[36] C. H. Peng, J. Scricco, S. Li, M. Fryd, B. B. Wayland, Macromole-cules 2008, 41, 2368 – 2373.

[37] Y. Champouret, U. Baisch, R. Poli, L. Tang, J. L. Conway, K. M.Smith, Angew. Chem. 2008, 120, 6158 – 6161; Angew. Chem. Int. Ed.2008, 47, 6069 –6072.

[38] J. Brandrup, E. H. Immergut, E. A. Grulke in Polymer Handbook,Wiley, New York, 1999.

[39] B. Smirnov, I. Bel’govskii, G. Ponomarev, A. Marchenko, N. Eniko-lopian, Dokl. Akad. Nauk SSSR 1980, 254, 127 – 130.

[40] A. A. Gridnev, Vysokomol. Soedin. Ser. A 1989, 31, 2153 –2159.[41] A. A. Gridnev, S. D. Ittel, Chem. Rev. 2001, 101, 3611 –3659.[42] A. Goto, Y. Kwak, T. Fukuda, S. Yamago, K. Iida, M. Nakajima, J.

Yoshida, J. Am. Chem. Soc. 2003, 125, 8720 –8721.[43] R. C. Elder, Inorg. Chem. 1968, 7, 1117 –1123.[44] A. Debuigne, C. Michaux, C. J�r�me, R. J�r�me, R. Poli, C. De-

trembleur, Chem. Eur. J. 2008, 14, 7623 –7637.[45] J. Harvey, M. Aschi, Faraday Discuss. 2003, 124, 129 – 143.[46] J. N. Harvey, Struct. Bonding (Berlin) 2004, 112, 151 –184.[47] F. A. Cotton, J. S. Wood, Inorg. Chem. 1964, 3, 245 –251.[48] B. K. Langlotz, J. Loret Fillol, J. H. Gross, H. Wadepohl, L. H.

Gade, Chem. Eur. J. 2008, 14, 10267 – 10279.[49] S. F. Boys, F. Bernardi, Mol. Phys. 1970, 19, 553 –566.[50] F. A. Cotton, C. Y. Liu, C. A. Murillo, X. P. Wang, Inorg. Chem.

2003, 42, 4619 –4623.[51] A. Altomare, M. Burla, M. Camalli, G. Cascarano, C. Giacovazzo,

A. Guagliardi, A. Moliterni, G. Polidori, R. Spagna, J. Appl. Crystal-logr. 1999, 32, 115 –119.

[52] G. M. Sheldrick, SHELXL97, Program for Crystal Structure refine-ment, University of Gçttingen, Gçttingen, Germany, 1997.

[53] L. J. Farrugia, J. Appl. Crystallogr. 1997, 30, 565.[54] Gaussian 03, Revision C.02, M. J. Frisch, H. B. Schlegel, G. E. Scuse-

ria, M. A. Robb, J. R. Cheeseman, J. Montgomery, Jr., J. A., T.Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyengar, J.Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega,G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R.Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao,H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J. B. Cross,C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev,A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala, K.Morokuma, G. A. Voth, P. Salvador, J. J. Dannenberg, V. G. Zakr-zewski, S. Dapprich, A. D. Daniels, M. C. Strain, O. Farkas, D. K.Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. V. Ortiz,Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G.Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox,T. Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, M. Challa-combe, P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, C. Gonza-lez, J. A. Pople, Gaussian, Inc., Wallingford, CT, 2004.

[55] A. D. Becke, J. Chem. Phys. 1993, 98, 5648 –5652.

www.chemeurj.org � 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2009, 15, 4874 – 48854884

R. Poli, Y. Gnanou et al.

[56] J. P. Perdew in Electronic structure of solids (Ed.: P. Ziesche, H. Es-chrig), Akademie Verlag, Berlin, 1991, pp. 11.

[57] K. Burke, J. P. Perdew, Y. Wang in Electronic Density FunctionalTheory: Recent Progress and New Directions (Eds.: J. F. Dobson, G.Vignale, M. P. Das), Plenum Press, New York, 1998.

[58] P. J. Hay, W. R. Wadt, J. Chem. Phys. 1985, 82, 270 –283.

Received: November 17, 2008Published online: March 23, 2009

Please note: Minor changes have been made to this publication inChemistry—A European Journal. The Editor.

Chem. Eur. J. 2009, 15, 4874 – 4885 � 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 4885

FULL PAPERCoexistence of Three Polymerization Mechanisms