Homolysis and Decomposition of Alkoxyamines Containing PROXYL and TEMPO Residues: A Comparison

Macromol. Chem. Phys. 2003, 204, 1923–1932 Full Paper: Quantitative ESR and 1H NMR spectroscopies have been used to study the homolysis and decomposition, respectively, of alkoxyamines derived from TEMPO and PROXYL nitroxides. It is shown that alkoxyamines substituted in the b position with a tert-butoxy group have different activation parameters than those bearing a 1-phenethyl fragment, however, there is compensation between transition state enthalpy and entropy, resulting in most cases in similar transition state free energies and homolysis kinetics. On the other hand, b-substituted alkoxyamines show a much lower tendency to undergo decomposition, while species derived from TEMPO are more prone to decomposition 1923 than those from the substituted PROXYL derivative investigated. These findings are used to explain the observed differences in polymerisation behaviour of the various alkoxyamines. Alkoxyamines used in the present study. Homolysis and Decomposition of Alkoxyamines Containing PROXYL and TEMPO Residues: A Comparison Catherine A. Bacon, Neil R. Cameron,* Alistair J. Reid Department of Chemistry, University of Durham, South Road, Durham, DH1 3LE, UK Fax: (þ44) 191 384 4737; E-mail: n.r.cameron@durham.ac.uk Received: May 13, 2003; Revised: August 18, 2003; Accepted: September 16, 2003; DOI: 10.1002/macp.200350074 Keywords: activation energy; alkoxyamines; ESR/EPR; nitroxides; thermodynamics Introduction Nitroxide-mediated polymerisation (NMP)[1] has emerged as one of several successful controlled radical polymerisation methods.[2] It is brought about by the dynamic equilibrium between dormant and active chains (Scheme 1), which is set up when radical polymerisations are conducted in the presence of a sufficient amount of nitroxide such as TEMPO (species on right hand side of reaction in Scheme 1). The low propagating radical concentration (in Scheme 1, K
1011) together with the Persistent Radical Effect[3] ensure that control is maintained during the reaction and narrow polydispersity products are obtained. In addition, functional architectures such as block/graft copolymers, telechelics and star polymers can be prepared, since the nitroxide residue remains covalently attached to the chain end. The most successful method of conducting NMP is to presynthesise an alkoxyamine initiator (species on the left hand side of reaction in Scheme 1) and add this to a charge of monomer,[4] in certain cases with a small quantity of free nitroxide to maintain control.[5] In this way, the exact concentration of initiating species is known Macromol. Chem. Phys. 2003, 204, No. 16 at the beginning of the reaction. However, the ability of any alkoxyamine to control efficiently the polymerisation of a given monomer depends on features such as its equilibrium constant for the reaction shown in Scheme 1 and whether or not any other reactions take place at an appreciable rate. Therefore, studies of the homolysis and decomposition of alkoxyamines are pertinent to understanding NMP processes. The usefulness of an alkoxyamine initiator requires it to have a short half-life so that all chains are initiated in a short timescale, while the kinetic scheme in operation for a given monomer/nitroxide system as set out by Fischer et al.,[3] either dominated by monomer autoinitiation or the persistent radical effect, is dependent on the equilibrium constant of the reversible capping reaction, K (¼ kd/kc), and the propagation rate constant at the polymerisation temperature. A number of studies of alkoxyamine homolysis have been performed. Moad and Rizzardo[6] determined the half-lives of several alkoxyamines in the presence of an excess of a different nitroxide as the trapping species to prevent the back reaction that regenerates the original alkoxyamine. It was shown that increased steric crowding in the DOI: 10.1002/macp.200350074 ß 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1924 C. A. Bacon, N. R. Cameron, A. J. Reid Scheme 1. NMP. Equilibrium between dormant and active chains in nitroxide segment led to a shortening of the corresponding alkoxyamine half-life (lower kd in Scheme 1). A semiempirical molecular orbital approach was utilized by Kazmaier et al.[7] to predict bond dissociation enthalpies for many alkoxyamines. These indicated that di-tert-butyl nitroxide (DBNO) would have a lower bond dissociation enthalpy than TEMPO. Subsequent experimental studies by Goto and Fukuda confirmed this.[8] Scaiano et al.[9] determined homolysis rates for a series of alkoxyamines with different transient radical fragments. The activation energies obtained suggest that replacing the benzylic hydrogens of the transient radical segment with methyl groups, increasing steric bulk and stability of the transient radical, leads to a reduction in the homolysis activation energy for a given nitroxide. The reverse reaction in Scheme 1, trapping of alkyl radicals by nitroxides, has also been studied. Bowry and Ingold[10] found that rate constants for trapping alkyl radicals by nitroxides decreased with increasing nitroxide steric bulk (lower kc in Scheme 1). These results were verified by Fischer et al.,[11] who investigated the cross-coupling between a wide range of nitroxides and alkyl radicals. Furthermore, it was stated by these authors that open chain nitroxides, which have a larger C–N–C angle than cyclic (5- and 6-membered) analogues and thus are more sterically compressed, have higher kd and lower kc values, which of course leads to significantly higher equilibrium constants than TEMPO and other cyclic nitroxides. Studies of the homolysis of polymeric alkoxyamines have produced varied results. German et al.[12] reported a large increase in the rate of TEMPO release, measured by quantitative ESR spectroscopy, from polystyryl alkoxyamines compared to low molecular weight species, which was largely ascribed to entropic effects. However, Fukuda and co-workers[8] studied the homolysis of polystyryl alkoxyamines terminated with TEMPO, DBNO and the phosphonato open chain nitroxide, DEPN, using a GPC resolution method, and observed only a two to three fold increase in kd over values for comparable small molecule alkoxyamines as determined by Fischer and co-workers.[13] Furthermore, Tordo et al.[14,15] studied the effect of chain length on homolysis of a polystyryl-DEPN adduct by quantitative ESR and found no clear influence of chain length on the rate of dissociation. However, a strong chain length dependence was observed for poly(butyl acrylate)-DEPN alkoxyamines;[15] kd for a 37 kg  mol1 Macromol. Chem. Phys. 2003, 204, No. 16 www.mcp-journal.de adduct was almost 6 times higher than the corresponding low molar mass analogues. In addition to homolysis, alkoxyamines can undergo side reactions such as decomposition to yield hydroxylamine and an unsaturated species derived from the alkyl fragment. Li et al.[16] investigated the decomposition of 2,20 ,6,60 -tetramethyl-1-(1-phenylethoxy)piperidine (STEMPO) by HPLC and 1H NMR. The thermal decomposition of this alkoxyamine resulted in the formation of styrene, 2,3-divinylbutene and ethylbenzene. Fukuda and co-workers[17] investigated the decomposition of an oligomeric alkoxyamine, PS-TEMPO, and found that the level of decomposition was significantly lower than that previously obtained for S-TEMPO. This reduction is lower than would be expected simply through a reduction in the number of b-hydrogens from three to two. Although the rate of decomposition was lower than that of S-TEMPO it was still significant; at temperatures comparable to those of polymerisation 10% decomposition of chain ends was observed in 10 h. In order to study a better analogue of the polystyrene chain end, Scaiano et al.[18] developed an alkoxyamine with a 1,3-diphenylpropyl fragment, which resembles closely the end of the growing chain. It was found that at 125 8C an activation/deactivation cycle for S-TEMPO has a one in thirty chance of leading to hydrogen transfer whereas no decomposition was noted with the species modelling the penultimate unit of the growing chain. The authors also speculated on the mechanism of decomposition and concluded that hydrogen transfer occurs via a concerted process. The reduced rate of decomposition for the penultimate unit alkoxyamine was ascribed to the presence of a significant energy barrier to rotation of the H–C–C–O bond to the presumed syn-periplanar geometry required for the concerted elimination. Goto et al.[19] determined by 1H NMR the rates of decomposition for several alkoxyamines based on TEMPO, 4-hydroxy-TEMPO (HTEMPO) and open chain nitroxides. The results showed that nitroxide structure was very significant, with the rate decreasing with increasing size of the nitroxide. It was noted that while the rates of activation for HTEMPO and TEMPO were similar, HTEMPO had a significantly lower rate of decomposition possibly due to a hindered conformation resulting from intramolecular hydrogen bonding in the nitroxide. The decomposition of an alkoxyamine analogous to a methacrylate chain end terminated by TEMPO was shown to be far more significant than that of the comparable styryl species. Examination of decomposition in the presence of oxygen as an alkyl radical trap showed a dramatic reduction in rate, suggesting that the majority of decomposition results from radical combination rather than from radicals generated within a solvent cage or via a concerted mechanism, contrary to the work of Scaiano et al.[18] Ananchenko and Fischer[20] also investigated the decomposition of low molar mass TEMPO-derived alkoxyamines. It was found that ß 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Homolysis and Decomposition of Alkoxyamines . . . the contribution of concerted elimination to the decomposition of S-TEMPO is minor, as concluded from experiments conducted in the presence of excess styrene in which decomposition was retarded. However, decomposition of a TEMPO-acrylate type alkoxyamine was shown to be less retarded in the presence of styrene, suggesting that a direct elimination mechanism makes a significant contribution (the authors discount cage effects). A mechanism to account for this was postulated. The same authors[21] and others[19] have also investigated decomposition of DEPN-based alkoxyamines. In contrast to TEMPO-containing species, little if any decomposition to alkene plus hydroxylamine was observed. This indicates a strong influence of nitroxide structure on propensity for decomposition. Thus, it is clear that decomposition to yield hydroxylamine and vinyl species can occur in NMP with certain mediators as a side reaction during activation-deactivation. Several authors have investigated the effect this can have on the outcome of polymerisations. Gridnev[22] showed that hydroxylamines formed by alkoxyamine decomposition are capable of retarding polymerisation through their oxidation to nitroxides by carbon centred radicals. This suggests that catalytic chain termination may take place. Yang et al. studied the effect of hydroxylamine formation and catalytic chain termination on kinetics and molecular weight distribution in controlled radical polymerisation through Monte Carlo simulations.[23] The inclusion of hydrogen transfer reactions into kinetic calculations resulted in a reduction in polymerisation rate and a cessation of the polymerisation. The calculated kinetic schemes were compared to real polymerisations of styrene and methacrylates. Styrene, although capable of undergoing decomposition, has a supply of radicals from autoinitiation; methacrylates are more susceptible to hydrogen transfer but are not capable of undergoing autoinitiation. Good agreement between experimental systems and simulations of polymerisation with and without autoinitiation was observed. When autoinitiation is included in the simulations, the system polymerises smoothly with polydispersity decreasing with conversion, while without autoinitiation, as in TEMPO mediated methacrylate polymerisation, the reaction ceases before reaching high conversion and the polydispersity passes through a minimum before increasing with conversion. It was concluded that hydrogen transfer reactions reduce the rate of polymerisation and result in a broadening of the polydispersity. It was suggested that the increased propensity of methacrylates towards hydrogen transfer and the absence of autoinitiation may be responsible for the problems in polymerising these monomers with TEMPO. Previously, we have studied[24–26] NMP involving PROXYL nitroxides, which are the 5-membered ring analogues of TEMPO. We chose to study these initially because their functionalization, particularly adjacent to the ring N atom, is more accessible synthetically than that of TEMPO. Macromol. Chem. Phys. 2003, 204, No. 16 www.mcp-journal.de 1925 Scheme 2. Alkoxyamines used in the present study. Quantitative comparison indicated that a PROXYL derivative bearing an a-Ph substituent was found to mediate styrene polymerisation at around 2.5 times the rate of TEMPO with no loss of control.[24] More recently, it has been shown that the same mediator can bring about the living (but not controlled) polymerisation of butyl acrylate, which cannot be achieved with TEMPO.[25,26] Therefore, we were keen to investigate the properties of alkoxyamines derived from 2-phenyl-PROXYL and TEMPO to elucidate the reasons for their different performances in NMP. The alkoxyamines chosen are those initiators that have been used previously in our NMP experiments (Scheme 2, 1–3) and are not meant to be accurate models of the growing polystyrene chain end. For comparison, we have also investigated analogous alkoxyamines bearing 1-phenethyl residues (Scheme 2, 4 and 5). Experimental Part Materials and Instrumentation Toluene (solvent for ESR studies) was distilled before use. TEMPO was purified by vacuum sublimation. The inhibitor was removed from styrene by passing through a short pad of basic Al2O3. 2,20 ,5-trimethyl-50 -phenylpyrrolidinyl-N-oxyl (TMPhNO)[27] and di-tert-butyl peroxalate (DTBPO)[28] were prepared by literature methods (Caution: DTBPO when dry is known to be shock sensitive and to detonate in contact with metal objects). All other chemicals and solvents were used as received from commercial suppliers (mainly Aldrich). ESR spectra were recorded on a Bruker EMX spectrometer fitted with a variable temperature probe. NMR spectra were obtained with a Varian Innova 400 fitted with a variable temperature probe, operating at either 400 MHz (1H) or 100 MHz (13C), using either CDCl3 or [D8]toluene as the solvent and tetramethylsilane (TMS) as an internal standard. Mass spectroscopy was performed with a Micromass Auto Spec. Alkoxyamine Synthesis Alkoxyamine 5 was prepared from TMPhNO according to Hawker’s procedure involving Jacobsen’s catalyst:[29] to a ß 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1926 C. A. Bacon, N. R. Cameron, A. J. Reid These were placed to the same depth as the previous standards in the ESR tube and the top of the tube was sealed. The ESR spectrometer was then tuned at room temperature following which the tube was removed while the spectrometer equilibrated thermally. At time zero the tube was placed in the spectrometer, retuned and spectra were obtained at regular intervals thereafter. Nitroxide stability was checked by heating a solution of each in distilled toluene for 12 h in the ESR cavity. The spectra showed no change over this period. Scheme 3. Numbering for NMR assignment of 5. solution of styrene (1.00 g, 9.6 mmol) and TMPhNO (1.06 g, 5.2 mmol) in 100 ml of EtOH was added (R,R)-()-N,N0 -bis(3,5-di-tert-butylsalicylidene)-1,2-cyclohexanediamino-manganese(III) chloride (Jacobsen’s catalyst, 1.10 g, 1.7 mmol) and NaBH4 (0.71 g, 19 mmol). The solution was allowed to stir overnight at room temperature, after which the solvent was removed on the rotary evaporator. The residue was partitioned between water and DCM (200:150 ml), the layers were separated and the aqueous layer was washed with DCM (3  100 ml). The combined organic washings were dried over MgSO4, filtered and evaporated to dryness. The solid residue was purified by flash chromatography (petroleum ether:isopropyl alcohol 99:1) yielding 0.47 g (29%) of 5. Refer to Scheme 3 for NMR assignments. C21H26NO (FW 308.445): Calcd. C 81.8, H 8.4, N 4.5; Found C 81.9; H, 8.7; N, 4.5. 1 H NMRa (400 MHz, CDCl3): d ¼ 0.67, 1.05, 1.10, 1.43, 1.50, 1.59 (s, H1, H2 and H7, 18H), 1.00, 1.37 (d, H13, 6H), 1.63–1.95 (m, H4 and H5, 8H), 4.21 (q, H12, 2H), 6.90–7.61 (m, H9-11 and H15-17, 10H). 13 C NMR (100 MHz, CDCl3): d ¼ 22.2, 22.3, 22.8, 22.9, 23.5, 23.6, 29.9, 30.0 (C1, C2, C7 and C13), 38.2, 38.3 (C4), 40.2, 40.3 (C5), 64.1, 64.3 (C3), 68.1, 68.2 (C6), 79.8, 81.3 (C12), 125–130 (C9-11 and C15-17), 144.0, 145.3 (C14), 152.7, 152.8 (C8). LRMS (EI): 309 [M þ 1], 205 [nitroxide þ 1], 190 [nitroxþ ide  Me þ 1], 131, 118, 105 [PhEtþ], 91 [PhCHþ 2 ], 77 [C6H5 ]. Compound 4 was synthesized according to a similar procedure; 1 to 3 were prepared from styrene, nitroxide and DTBPO according to a literature procedure.[24] In all cases, characterisation data were in accord with literature values. ESR Studies Standard solutions of TEMPO, PROXYL and TMPhNO in distilled toluene were prepared for the construction of calibration plots. These were placed to a constant depth in an ESR tube. The top of the tube was then sealed with plastic film to minimize evaporation. The tube was placed in the ESR spectrometer and allowed to equilibrate thermally and the ESR signal was measured. Non-degassed solutions of the five alkoxyamines under investigation, 1 to 5, were then prepared. a Note that 5 exists as a mixture of diastereomers. Macromol. Chem. Phys. 2003, 204, No. 16 www.mcp-journal.de NMR Studies Alkoxyamine was dissolved in [D8]toluene and the resulting solution was placed in an NMR tube fitted with a Young’s tap. This was degassed by three freeze-pump-thaw cycles and sealed under vacuum. The tube was then placed in the cavity of a NMR spectrometer with variable temperature probe, previously equilibrated at 125 8C. 1H NMR spectra were recorded hourly. The quantity of styrene evolved from 4 over time was determined from the NMR spectra (see Figure 3a). The integral of the peak at d ¼ 4.75 (q, CH3CH) gave the area per proton of the alkoxyamine component (Aa) while the average of the integrals of the peaks at d ¼ 5.0, 5.5 and 6.5 gave the area per proton due to styrene (As). The conversion of alkoxyamine to styrene could then be determined from As/(Aa þ As). Results and Discussion Alkoxyamine Homolysis The most direct method of studying alkoxyamine homolysis is electron spin resonance (ESR), which gives a direct measurement of nitroxide formation without the need of other approaches for offline sampling. Fischer et al.[13] performed quantitative ESR measurements, in the presence of excess trapping species, on a series of 27 alkoxyamines based on six nitroxides with different carbon centred radical fragments. The bond strengths were calculated from Arrhenius plots constructed from the alkoxyamine homolysis rate constants over a range of temperatures. It was suggested that the observed variation of rate constants was not due to changes in Arrhenius frequency factor (A) but was the result of bond strength differences. The frequency factors observed were similar to those obtained previously by Scaiano et al.;[18] the systems evaluated had an average value of A ¼ 2.6  1014 s1 with a 2.5 fold variation about this average. ESR spectra were collected from alkoxyamine samples 1 to 5 cleaved in the presence of oxygen as a radical scavenger at different temperatures,[12] resulting in an increasing double integral of ESR signal with time (Figure 1). To determine homolysis rate constants from these data, a correlation between double integral of ESR signal and nitroxide concentration was established by means of standard solutions of the nitroxides in question. Subsequently, a linear fit to the double integral with time data allowed the calculation of alkoxyamine bond homolysis rate constants ß 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Homolysis and Decomposition of Alkoxyamines . . . 1927 Figure 1. Evolution of the double integral of the ESR signal with time for 5 at: 330 K (^); 340 K (&); 350 K (~); 360 K (*); and 370 K (). (kd) at each temperature studied, which led to an Arrhenius plot (ln(kd) vs. 1/T) for each alkoxyamine (Figure 2). In each case, a nonlinear least squares fitting algorithm was used to obtain the activation energy Ea and the pre-exponential factor A, from the slope and intercept respectively. From Figure 2 it can be seen that all alkoxyamines have similar homolysis rate constants at each temperature, apart from 2. It has been reported elsewhere[6] that alkoxyamine C–O bond dissociation energy decreases as nitroxide fragment ring size increases from 5 to 6, which is in line with the lower kd values of 2 compared to 1. The higher rate constants for 3 compared to 2 are ascribed to the greater steric bulk of the former.[6,13] The activation thermodynamics for each alkoxyamine were then determined using a standard transition state approach. The Eyring equation kd ¼ kB T DH z =RT DSz =R e e h ð1Þ where kB is the Boltzmann constant, h is Planck’s constant, R is the gas constant and DH{ and DS{ are the enthalpy and entropy of activation, respectively, can be rewritten as:   kd h DH z DSz þ ð2Þ ¼ ln RT R kB T Table 1. A plot of the left hand side of (2) versus 1/T gives DH{ from the gradient and DS{ from the intercept. Knowing these allows the transition state free energy to be calculated from the Gibbs equation DGz ¼ DH z  TDSz ð3Þ The activation parameters for alkoxyamines 1 to 5 are shown in Table 1. Given the differences in polymerisation behaviour between the alkoxyamines, it is surprising that most alkoxyamines have similar homolysis kinetics. However, a close analysis of the activation parameters (Table 1) reveals some significant differences amongst alkoxyamines. Focusing in the first instance on the Arrhenius parameters (Ea and A), it can be seen that the alkoxyamines fall into two classes distinguished by their alkyl fragments. Those with 1-phenethyl residues (4 and 5) have significantly higher values of Ea and A than alkoxyamines derived from tert-butoxy radicals (1 to 3). Evidently, the presence of the b-alkoxy fragment has a significant effect on alkoxyamine homolysis. The Activation parameters for alkoxyamines investigated. Alkoxyamine 1 2 3 4 5 a) Figure 2. Arrhenius plots for alkoxyamines: 1 (^); 2 (&); 3 (~); 4 (); and 5 (*). A Ea DH{ DS{ DG{a) s1 kJ  mol1 kJ  mol1 J  K1  mol1 kJ  mol1 3.2  1010 2.0  109 2.0  1011 3.2  1014 1.6  1013 102.5 104.3 109.8 129.9 121.8 99.6 101.5 106.9 127.0 118.9 53.5 78.4 38.0 23.1 1.7 118.3 128.9 120.2 118.9 119.5 At 350 K. Macromol. Chem. Phys. 2003, 204, No. 16 www.mcp-journal.de ß 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1928 C. A. Bacon, N. R. Cameron, A. J. Reid Scheme 4. Alkoxyamine transition state conformation leading to greatest stabilisation of developing radical by neighbouring C–O s* orbital. sions from the Arrhenius data. The greater steric bulk of the alkoxy fragment will also impose a significant restriction to rotation about the C–C bond of the alkoxy group, leading to a further decrease of DS{. Steric effects also provide an explanation for the difference between 4 and 5; the phenyl group on the nitroxide fragment produces greater steric congestion, leading to lower values of both DH{ and DS{. Decomposition reduction of Ea may arise from the stabilization of the developing alkyl radical by the b-alkoxy substituent via an anomeric effect (partial delocalisation of the unpaired electron into the alkoxy C–O s* orbital)[30] as shown in Scheme 4, although steric effects are also likely to be important. The A factors for 1 to 3 are significantly lower than values reported elsewhere for 1-phenethyl derived alkoxyamines. This may be due to restricted rotation around the alkoxyamine O–C–C–O bond in the transition state; the molecule will have to adopt a conformation where the alkoxy fragment C–O bond and the breaking alkoxyamine C–O bond are antiperiplanar to enable the anomeric stabilization. The A value for 1 is also much lower than those reported for the same species by Bon et al.[12] At present we have no explanation for this difference, but have confidence in our own data based on the agreement between the values of A and Ea for 4 with those reported in the literature.[13] Ea for 5 is also in line with reported values for other 1-phenethyl derived alkoxyamines, although A is around one order of magnitude lower.[13] The differences in A values observed among the series of alkoxyamines examined are accompanied by similar variations in Ea. Thus, a lower value of Ea is always observed alongside lower A. The effect of this is that all alkoxyamines apart from 2 have similar kd values. To elucidate further the different behaviour of the alkoxyamines investigated, the activation thermodynamics were investigated. Compensation between enthalpy and entropy of activation results in similar DG{ values for alkoxyamines 1 and 3 to 5 (Table 1). Once again, the only substantially different species in this respect is 2, which has a significantly higher DG{ value than the others. This is in agreement with the kinetic data. However, as before a more detailed inspection of the data reveals that there are differences between alkoxyamines depending on the type of alkyl fragment. 1 to 3 have lower activation enthalpies than 4 and 5, but this is accompanied by lower (less favourable) activation entropies. A possible explanation, as before, is that the presence of the alkoxy group in 1 to 3 stabilises the developing radical and results in increased steric congestion in the transition state, leading to a lowering of bond dissociation energy (DH{ is roughly equivalent to BDE). However, the decrease in DS{ indicates that there is restriction to rotation about bonds in the transition state, which agrees with the concluMacromol. Chem. Phys. 2003, 204, No. 16 www.mcp-journal.de The ESR data alone do not provide an explanation for the differences in polymerisation behaviour of the various alkoxyamines, namely that 3 mediates the polymerisation of styrene significantly faster than that of both 1[24] and 5b together with its ability to bring about butyl acrylate polymerisation.[26] In an attempt to rationalize these observations, we examined the decomposition of various alkoxyamines to generate unsaturated species and hydroxylamine (see Scheme 5). Heating 4 at 125 8C for a period of time resulted in the formation of significant quantities styrene, as determined by 1H NMR spectroscopy. Figure 3a shows the spectrum obtained after 9 h and styrene resonances are clearly seen at d ¼ 5.0, 5.5 and 6.5. The presence of hydroxylamine was not verified but is expected based on the findings of others.[20] Furthermore, it is assumed that styrene is produced by the route shown in Scheme 5 rather than by a disproportionation reaction between 1-phenethyl radicals. We believe this is a reasonable assumption based on the lack of evidence for the latter reaction, e.g. ethylbenzene, found by Ananchenko et al. during a study of the decomposition of the same alkoxyamine.[20] The method of Fukuda et al.[19] was used to determine the decomposition rate constant kdec for 4. Assuming that self-termination and disproportionation reactions between alkyl radicals are unimportant, decomposition can be assumed to be first order in alkoxyamine concentration,[17] i.e. kdec is the pseudo-first order decomposition rate constant. The validity of the first order assumption was verified by Ohno et al.,[17] who found that the addition of excess TEMPO had no influence on the kinetics of PS-TEMPO decomposition. It then follows that lnðC0 =CÞ ¼ kdec t ð4Þ where C0 and C are the initial and actual concentrations of alkoxyamine. C0 is proportional to the sum of the areas per proton for alkoxyamine and styrene at time t and similarly C is related to the alkoxyamine integral, therefore kdec can be determined from the slope of the plot of ln([(Aa þ As)/Aa] (see Experimental for definition of terms) against time. A b Data not shown. ß 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Homolysis and Decomposition of Alkoxyamines . . . 1929 Figure 3. 1H NMR spectra of TEMPO-derived alkoxyamines heated in [D8]toluene at 125 8C for different times: (a) 4 after 9 h; (b) 1 after 16 h. Scheme 5. good linear fit to the data was obtained (Figure 4: R2 ¼ 0.9995), giving a value of kdec of 4.8  106 s1. This is compared with the value of 4.5  105 s1 determined by Fukuda and co-workers for the same alkoxyamine at 140 8C. In contrast, heating related alkoxyamine 1 resulted in the production of very low levels of vinyl containing species even after 16 h (spectra were recorded every hour). The level of unsaturated species was too low to obtain a reliable integration (Figure 3b; vinyl resonance circled). This indicates that the introduction of a b-substituent greatly reduces the propensity for TEMPO-derived alkoxyamines to undergo decomposition, in agreement with findings by other Decomposition of alkoxyamine 4. Macromol. Chem. Phys. 2003, 204, No. 16 www.mcp-journal.de ß 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1930 C. A. Bacon, N. R. Cameron, A. J. Reid Figure 4. Kinetic plot for the evolution of styrene from 4 (line shows best linear fit to data points). workers.[17–19] Alkoxyamines 3 and 5, based on TMPhNO, showed much less formation of unsaturated species over long periods of time than corresponding TEMPO-derived analogues 1 and 4 (Figure 5). The 1-phenethyl alkoxyamine leads to the production of observable quantities of styrene (Figure 5a; styrene resonances circled) after 16 h at 125 8C (again, spectra were recorded hourly), however, not enough to be quantified reliably and certainly much less than analogous alkoxyamine 4. This demonstrates the powerful influence of nitroxide residue structure on the tendency of alkoxyamines to undergo decomposition. Furthermore, b-tert-butoxy substituted derivative 3 under the same conditions does not generate unsaturated decomposition products at NMR-detectable levels (Figure 5b). These results tell us two things: b-substitution reduces the tendency for decomposition of PROXYL containing alkoxyamines Figure 5. 1H NMR spectra of TMPhNO-derived alkoxyamines heated in [D8]toluene at 125 8C for 16 h: (a) 5; (b) 3. Macromol. Chem. Phys. 2003, 204, No. 16 www.mcp-journal.de ß 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Homolysis and Decomposition of Alkoxyamines . . . (compare Figure 5b with 5a); and PROXYL-derived alkoxyamines are less prone to decomposition than those based on TEMPO (compare Figure 5a with 4a). The lower tendency of b-alkoxy alkoxyamines to undergo decomposition could simply be due to the lower number of b-hydrogens compared to 1-phenethyl derivatives, however it is likely that the b-substituent plays a more significant role. Scaiano et al.[18] found that a TEMPOderived alkoxyamine possessing an alkyl fragment with a b-substituent did not undergo decomposition, whereas 4 did. Three possible mechanisms for the decomposition of 4 (and similar species) were considered: out-of-cage disproportionation; in-cage disproportionation; and concerted elimination. Experiments conducted in the presence of a large excess of a scavenging nitroxide indicated that cross-disproportionation, i.e. between alkyl radicals and the scavenger, did not occur, on the basis of which the authors ruled out the first of the above mechanisms. Consideration of the rate constants for recombination and diffusion lead the authors to conclude that the second mechanism was highly unlikely (less than 1% of encounters between alkyl radical and nitroxide). From this the authors then concluded that decomposition occurred via concerted elimination. To explain the lack of decomposition observed with their b-substituted alkoxyamine, they considered the barrier to rotation required to achieve the (presumed) correct conformation for concerted elimination, i.e. a syn-periplanar arrangement of the H–C–C–O species. The b-substituted alkoxyamine obviously has a greater rotational barrier than species 4. Subsequent work by Fukuda and co-workers[19] however casts some doubt on these conclusions. These authors observe that decomposition is severely reduced in the presence of oxygen (rather than another nitroxide) as a scavenger. This leads to the conclusion that concerted elimination is an unimportant, or at best a minor, pathway. Fukuda et al. conclude that decomposition occurs during the deactivation cycle (recombination between alkyl radical and nitroxide). We are of the opinion that the experiments of Fukuda et al. conducted in the presence of oxygen, which would be extremely unlikely to abstract hydrogen from an alkyl radical, provide strong evidence against the concerted elimination mechanism. Relevance to Use in Polymerisations We can relate the results described above to our observations of the polymerisation behaviour of PROXYL-derived alkoxyamines. Previously, we found that 3 gave rise to more rapid styrene polymerisation than 1. The similarity of their activation parameters (Table 1) suggests that decomposition is the dominant factor in this case. The greater tendency of 1 to decompose generates a small amount of hydroxylamine, which can transfer its hydrogen to a propagating radical, generating the corresponding nitroxide plus a dead chain. The effect of this is to cause slight reMacromol. Chem. Phys. 2003, 204, No. 16 www.mcp-journal.de 1931 tardation of the polymerisation (lower radical and greater nitroxide concentration). Of course, the decomposition results discussed above are only relevant to the initiator homolysis phase; nonetheless, the difference between 1 and 3 is due to the nitroxide fragment, so it should persist during the polymerisation leading to a weak catalytic chain transfer process (as observed by Gridnev).[22] We have also observed that the polymerisation of styrene mediated by 5 (results not shown) is significantly slower than that with 3. From Figure 2 and Table 1 it can be seen that 5 and 3 have similar kd values and free energies of activation, but 5 is more prone to b-hydrogen abstraction than 3 (Figure 5a). Thus, the difference in polymerisation rate is also likely to be due to hydrogen abstraction proficiency. In this case, the difference lies in the alkyl fragment, so it will only be apparent at the initiation of each chain. Despite this, decomposition during initiation will result in quantities of hydroxylamine that will have a retardative effect. Differences in ability to polymerise butyl acrylate have also been found. 3 gives rise to a rapid, living but uncontrolled polymerisation whereas both 1 and 5 are inactive. Since all have similar activation parameters (Table 1) we conclude again that the lower tendency of 3 to undergo decomposition is governing its behaviour. It is presumed that the reduced propagating radical and increased nitroxide concentration resulting from the generation of small quantities of hydroxylamine is sufficient to kill the polymerisation of butyl acrylate with 1 and 5 (particularly as this monomer does not autoinitiate). Recent work from Studer’s group[31] has shown that hydrogen abstraction is also the dominant factor in dictating the behaviour of alkoxyamines derived from substituted 7- and 8-membered ring nitroxides. Conclusion The homolysis and decomposition of alkoxyamines containing PROXYL residues have been investigated and compared to the behaviour of analogous species derived from TEMPO. It is shown that the alkoxyamine structure influences the individual activation parameters (Ea, A, DH{ and DS{), and that it is the alkyl rather than the nitroxide fragment that has the greatest influence. However, there is compensation between transition state enthalpy and entropy, resulting in most alkoxyamines having similar DG{ values and homolysis kinetics. It is suggested that the b-alkoxy substituent aids dissociation by stabilising partially the developing radical in the transition state, through interaction with the neighbouring C–O s* orbital (anomeric effect), although steric congestion in the transition state cannot be discounted. However, restriction to rotation in the transition state is also produced, lowering DS{ at the same time. Decomposition of the alkoxyamines was also studied by 1H NMR spectroscopy. It was found that ß 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1932 C. A. Bacon, N. R. Cameron, A. J. Reid b-substitution greatly reduced the tendency of an alkoxyamine to undergo decomposition to generate unsaturated species, and that TEMPO-derived alkoxyamines gave rise to more decomposition than those obtained from PROXYLs. These results indicate that differences in polymerisation behaviour between alkoxyamines derived from these cyclic nitroxides are due mainly to variations in their tendency to undergo decomposition. Acknowledgement: The authors would like to thank the EPSRC and Schlumberger for funding this work, Dr. M.R. Crampton (University of Durham) and Dr. G. Tustin (Schlumberger Cambridge Research) for helpful discussions and Dr. A. Royston (University of Durham) for assistance with ESR experiments. The EPSRC is also gratefully acknowledged for a Joint Infrastructure Fund grant (GR/M87917) with which the ESR spectrometer was purchased. [1] [1a] M. K. Georges, R. P. N. Veregin, P. M. Kazmaier, G. K. Hamer, Macromolecules 1993, 26, 2987; [1b] C. J. Hawker, A. W. Bosman, E. Harth, Chem. Rev. 2001, 101, 3661. [2] [2a] ‘‘Controlled Radical Polymerization’’, K. Matyjaszewski, Ed.; ACS Symp. Ser. 1998, 685, Am. Chem. Soc., Washington, DC 1998; [2b] ‘‘Controlled/Living Radical Polymerization’’, K. Matyjaszewski, Ed.; ACS Symp. Ser. 2000, 768, Am. Chem. Soc., Washington, DC 2000; [2c] ‘‘Advances in Controlled/Living Radical Polymerization’’, K. Matyjaszewski, Ed.; ACS Symp. Ser. 2003, 854, Am. Chem. Soc., Washington, DC 2003. [3] H. Fischer, Chem. Rev. 2001, 101, 3581. [4] C. J. Hawker, G. G. Barclay, A. Orellana, J. Dao, W. Devonport, Macromolecules 1996, 29, 5245. [5] [5a] D. Benoit, V. Chaplinski, R. Braslau, C. J. Hawker, J. Am. Chem. Soc. 1999, 121, 3904; [5b] P. LacroixDesmazes, J. F. Lutz, F. Chauvin, R. Severac, B. Boutevin, Macromolecules 2001, 34, 8866. [6] G. Moad, E. Rizzardo, Macromolecules 1995, 28, 8722. [7] P. M. Kazmaier, K. A. Moffat, M. K. Georges, R. P. N. Veregin, G. K. Hamer, Macromolecules 1995, 28, 1841. [8] A. Goto, T. Fukuda, Macromol. Chem. Phys. 2000, 201, 2138. Macromol. Chem. Phys. 2003, 204, No. 16 www.mcp-journal.de [9] W. G. Skene, S. T. Belt, T. J. Connolly, P. Hahn, J. C. Scaiano, Macromolecules 1998, 31, 9103. [10] V. W. Bowry, K. U. Ingold, J. Am. Chem. Soc. 1992, 114, 4992. [11] J. Sobek, R. Martschke, H. Fischer, J. Am. Chem. Soc. 2001, 123, 2849. [12] S. A. F. Bon, G. Chambard, A. L. German, Macromolecules 1999, 32, 8269. [13] S. Marque, C. Le Mercier, P. Tordo, H. Fischer, Macromolecules 2000, 33, 4403. [14] D. Bertin, F. Chauvin, S. Marque, P. Tordo, Macromolecules 2002, 35, 3790. [15] O. Guerret, J. L. Couturier, F. Chauvin, H. El-Bouazzy, D. Bertin, D. Gigmes, S. Marque, H. Fischer, P. Tordo, ACS Symp. Ser. 2003, 854, 412. [16] I. Li, B. A. Howell, K. Matyjaszewski, T. Shigemoto, P. B. Smith, D. B. Priddy, Macromolecules 1995, 28, 6692. [17] K. Ohno, Y. Tsujii, T. Fukuda, Macromolecules 1997, 30, 2503. [18] W. G. Skene, J. C. Scaiano, P. A. Yap, Macromolecules 2000, 33, 3536. [19] A. Goto, Y. Kwak, C. Yoshikawa, Y. Tsujii, Y. Sugiura, T. Fukuda, Macromolecules 2002, 35, 3520. [20] G. S. Ananchenko, H. Fischer, J. Polym. Sci., Part A: Polym. Chem. 2001, 39, 3604. [21] G. S. Ananchenko, M. Souaille, H. Fischer, C. Le Mercier, P. Tordo, J. Polym. Sci., Part A: Polym. Chem. 2002, 40, 3264. [22] A. A. Gridnev, Macromolecules 1997, 30, 7651. [23] [23a] J. He, H. Zhang, J. Chen, Y. Yang, Macromolecules 1997, 30, 8010; [23b] J. P. He, L. Li, Y. L. Yang, Macromolecules 2000, 33, 2286; [23c] J. He, L. Li, Y. L. Yang, Macromol. Theory Simul. 2000, 9, 463. [24] N. R. Cameron, A. J. Reid, P. Span, S. A. F. Bon, J. J. G. S. van Es, A. L. German, Macromol. Chem. Phys. 2000, 201, 2510. [25] N. R. Cameron, A. J. Reid, Polym. Prepr. (Am. Chem. Soc., Div. Polym. Chem.) 2002, 43, 88. [26] N. R. Cameron, A. J. Reid, Macromolecules 2002, 35, 9890. [27] [27a] J. F. W. Keana, S. E. Seyedrezai, G. Gaughan, J. Org. Chem. 1983, 48, 2644; [27b] K. Hideg, L. Lex, J. Chem. Soc.,Chem. Commun. 1984, 1263; [27c] K. Hideg, H. O. Hankovszky, H. A. Halasz, P. Sohar, J. Chem. Soc., Perkin Trans. 1 1988, 2905; [27d] W. R. Couet, R. C. Brasch, G. Sosnovsky, J. Lukszo, I. Prakash, C. T. Gnewuch, T. N. Tozer, Tetrahedron 1985, 41, 1165. [28] P. D. Bartlett, E. P. Benzing, R. E. Pincock, J. Am. Chem. Soc. 1960, 82, 1762. [29] J. Dao, D. Benoit, C. J. Hawker, J. Polym. Sci., Part A: Polym. Chem. 1998, 36, 2161. [30] J. F. King, R. Rathore, Z. R. Guo, M. Q. Li, N. C. Payne, J. Am. Chem. Soc. 2000, 122, 10308. [31] T. Schulte, A. Studer, Macromolecules 2003, 36, 3078. ß 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim