Atom Transfer Radical Polymerization Krzysztof Matyjaszewski* and Jianhui Xia Department of Chemistry, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213 Received February 15, 2001 Contents I. Introduction 2921 II. Mechanistic Understandings of Atom Transfer Radical Polymerization 2923 A. Components 2923 1. Monomers 2923 2. Initiators 2924 3. Catalysts 2924 4. Solvents 2924 5. Temperature and Reaction Time 2925 6. Additives 2925 B. Typical Phenomenology 2925 1. Kinetics 2925 2. Molecular Weight 2926 3. Molecular Weight Distribution 2926 4. Normal and Reverse ATRP 2927 5. Experimental Setup 2927 6. Catalyst Homogeneity 2927 7. Summary and Outlook 2928 C. ATRP Monomers 2928 1. Styrenes 2928 2. Acrylates 2929 3. Methacrylates 2929 4. Acrylonitrile 2930 5. (Meth)acrylamides 2930 6. (Meth)acrylic Acids 2931 7. Miscellaneous Monomers 2931 8. Summary and Outlook 2931 D. ATRP Initiators 2932 1. Halogenated Alkanes 2932 2. Benzylic Halides 2932 3. R-Haloesters 2933 4. R-Haloketones 2933 5. R-Halonitriles 2934 6. Sulfonyl Halides 2934 7. General Comments on the Initiator Structure in ATRP 2934 8. Summary and Outlook 2935 E. Transition-Metal Complexes 2935 1. Group 6: Molybdenum and Chromium 2935 2. Group 7: Rhenium 2936 3. Group 8: Ruthenium and Iron 2936 4. Group 9: Rhodium 2938 5. Group 10: Nickel and Palladium 2938 6. Group 11: Copper 2939 7. Summary and Outlook 2940 F. Ligand 2941 1. Nitrogen Ligands 2941 2. Phosphorus Ligands 2941 3. Miscellaneous Ligands 2942 4. Summary and Outlook 2942 G. Additives 2942 H. Catalyst Structure 2943 I. Mechanism 2945 J. Overall Elementary Reactions 2947 III. Materials Made by ATRP 2949 A. Functionality 2949 1. Monomer Functionality 2949 2. Initiator Functionality 2952 3. Chain End Functionality 2955 4. Summary and Outlook 2957 B. Composition 2957 1. Gradient/Statistical Copolymers 2958 2. Block Copolymers 2960 3. Inorganic/Organic Hybrids 2969 4. Summary and Outlook 2970 C. Topology 2972 1. Graft Copolymers 2972 2. Grafts from Surfaces 2977 3. Star Polymers 2978 4. Hyperbranched Polymers 2981 5. Summary and Outlook 2983 IV. Conclusions 2983 V. Acknowledgment 2985 VI. References 2985 I. Introduction The synthesis of polymers with well-defined com- positions, architectures, and functionalities has long been of great interest in polymer chemistry. Typi- cally, living polymerization techniques are employed where the polymerizations proceed in the absence of irreversible chain transfer and chain termination.1-3 Much of the academic and industrial research on living polymerization has focused on anionic, cationic, coordination, and ring-opening polymerizations. The development of controlled/living radical polymeriza- tion (CRP) methods has been a long-standing goal in polymer chemistry, as a radical process is more tolerant of functional groups and impurities and is the leading industrial method to produce polymers.4 Despite its tremendous industrial utility, CRP has not been realized until recently, largely due to the inevitable, near diffusion-controlled bimolecular radi- cal coupling and disproportionation reactions. 2921 Chem. Rev. 2001, 101, 2921-2990 10.1021/cr940534g CCC: $36.00 �� 2001 American Chemical Society Published on Web 09/12/2001

The past few years have witnessed the rapid growth in the development and understanding of new CRP methods.5,6 All of these methods are based on establishing a rapid dynamic equilibration between a minute amount of growing free radicals and a large majority of the dormant species. The dormant chains may be alkyl halides, as in atom transfer radical polymerization (ATRP) or degenerative transfer (DT), thioesters, as in reversible addition fragmentation chain transfer processes (RAFT), alkoxyamines, as in nitroxide mediated polymerization (NMP) or stable free radical polymerization (SFRP), and potentially even organometallic species. Free radicals may be generated by the spontaneous thermal process (NMP, SFRP) via a catalyzed reaction (ATRP) or reversibly via the degenerative exchange process with dormant species (DT, RAFT). All of the CRP methods, shown in Scheme 1, include activation and deactivation steps (with rate constants kact and kdeact), although in RAFT and DT the scheme may be formally simplified to just the exchange process with the apparent rate constant kexch. Generated free radicals propagate and termi- nate (with rate constants kp and kt), as in a conven- tional free-radical polymerization. Thus, although termination occurs, under appropriate conditions its contribution will be small (less than a few percent of total number of chains) and these radical polymer- izations behave as nearly living or controlled systems. This review will focus on the fundamentals of transition metal catalyzed atom transfer radical polymerization (ATRP). We will discuss the current mechanistic understanding of this process and some synthetic applications that have resulted in a variety of well-defined materials. This review covers the literature from the beginning of this field (1995) until approximately the end of 2000. We primarily refer to papers published in peer-reviewed journals, unless the work appeared in nonpeer-reviewed literature and was not followed by a full publication. A general mechanism for ATRP shown in Scheme 2 corresponds to case 2 from Scheme 1. The radicals, or the active species, are generated through a revers- ible redox process catalyzed by a transition metal Krzysztof (Kris) Matyjaszewski was born in Konstantynow, Poland, in 1950. He obtained his Ph.D. degree in 1976 at the Polish Academy of Sciences in Lodz, Poland, working in the laboratories of Professor S. Penczek. He has received his Habilitation Degree in 1985 from Lodz Polytechnic, Poland. He stayed as a postdoctoral fellow at the University of Florida, working with Professor G. B. Butler. Since 1985 he has been at Carnegie Mellon University, where he has served as Chemistry Department Head (1994-1998) and is currently J. C. Warner Professor of Natural Sciences. He is also an adjunct professor at the Department of Petroleum and Chemical Engineering at the University of Pittsburgh and the Polish Academy of Sciences in Lodz, Poland. He served as Visiting Professor at the Universities in Paris, Strasbourg, Bordeaux, Bayreuth, Freiburg, Ulm, and Pisa. He is an editor of Progress in Polymer Science and serves on seven editorial boards of polymer journals. His main research interests include controlled/living polymerization with the most recent emphasis on free-radical systems. In 1995 he developed atom transfer radical polymerization (ATRP), one of the most successful methods for controlled/ living radical polymerization (CRP) systems. During the last 5 years his group (25 postdoctoral fellows and 23 graduate and 26 undergraduate students) has published over 200 papers on ATRP and CRP. He holds over 20 U.S. and international patents. Close industrial interactions have been maintained by the ATRP Consortium (13 companies in 1996-2000) and newly established CRP Consortium (19 companies in 2001-2005). Research of Matyjaszewski group has received wide recognition, as evidenced by the ACS Carl S. Marvel Award for Creative Polymer Chemistry (1995), Elf Chair of French Academy of Sciences (1998), Humboldt Award for Senior US Scientists (1999), National Professorship of Poland (2000), Fellowship of ACS Division of Polymeric Materials and Engineering (2001), ACS Pittsburgh Award (2001), and ACS Award in Polymer Chemistry (2001). Jianhui Xia is a Senior Research Scientist in Corporate R&D at 3M Company in Saint Paul, MN. He received his B.S. degree in Polymer Chemistry in 1991 from the University of Science and Technology of China working with Professor Dezhu Ma. He then went to Emory University of Atlanta, GA, where he obtained his M.S. degree in Organic Chemistry on asymmetric synthesis from Professor Dennis Liotta. He earned his Ph.D. degree in Polymer Chemistry in 1999 on controlled/���living��� radical polymerization at Carnegie Mellon University under the direction of Professor Krzysztof Matyjaszewski. His current research interests include the controlled synthesis of novel polymeric materials. Scheme 1. General Scheme of CRP Methods 2922 Chemical Reviews, 2001, Vol. 101, No. 9 Matyjaszewski and Xia

complex (Mtn-Y/Ligand, where Y may be another ligand or the counterion) which undergoes a one- electron oxidation with concomitant abstraction of a (pseudo)halogen atom, X, from a dormant species, R-X. This process occurs with a rate constant of activation, kact, and deactivation kdeact. Polymer chains grow by the addition of the intermediate radicals to monomers in a manner similar to a conventional radical polymerization, with the rate constant of propagation kp. Termination reactions (kt) also occur in ATRP, mainly through radical coupling and dis- proportionation however, in a well-controlled ATRP, no more than a few percent of the polymer chains undergo termination. Other side reactions may ad- ditionally limit the achievable molecular weights. Typically, no more than 5% of the total growing polymer chains terminate during the initial, short, nonstationary stage of the polymerization. This pro- cess generates oxidized metal complexes, X-Mtn+1, as persistent radicals to reduce the stationary con- centration of growing radicals and thereby minimize the contribution of termination.7 A successful ATRP will have not only a small contribution of terminated chains, but also a uniform growth of all the chains, which is accomplished through fast initiation and rapid reversible deactivation. The name atom transfer radical polymerization (ATRP) originates from the atom transfer step, which is the key elementary reaction responsible for the uniform growth of the polymeric chains, in the same way that the addition-fragmentation is the key step in the RAFT process. ATRP has its roots in atom transfer radical addition (ATRA), which targets the formation of 1:1 adducts of alkyl halides and alkenes, also catalyzed by transition metal complexes.8 ATRA is a modification of Kharasch addition reaction, which usually occurs in the presence of light or conventional radical initiators.9 Because of the involvement of transition metals in the activation and deactivation steps, chemo-, regio-, and stereoselectivities in ATRA and the Kharasch addition may be different. For example, under Kharasch conditions, in the reaction with chloroform the alkene will ���insert��� across the H-CCl3 bond but in ATRA it will insert across the Cl-CHCl2 bond, because the C-Cl bond is rapidly activated by the Fe(II) or Cu(I) complexes.10 ATRP also has roots in the transition metal catalyzed telomerization reactions.11 These reactions, however, do not proceed with efficient exchange, which results in a nonlinear evolution of the molec- ular weights with conversions and polymers with high polydispersities. ATRP also has connections to the transition metal initiated redox processes as well as inhibition with transition metal compounds.12-14 These two techniques allow for either an activation or deactivation process, however, without efficient reversibility. ATRP was developed by designing an appropriate catalyst (transition metal compound and ligands), using an initiator with the suitable struc- ture, and adjusting the polymerization conditions such that the molecular weights increased linearly with conversion and the polydispersities were typical of a living process.15-19 This allowed for an unprec- edented control over the chain topology (stars, combs, branched), the composition (block, gradient, alternat- ing, statistical), and the end functionality for a large range of radically polymerizable monomers.17,20-24 Earlier attempts with heterogeneous catalyst and inefficient initiators were less successful.25 ATRP is among the most rapidly developing areas of chemistry, with the number of publications ap- proximately doubling each year. According to Sci- Finder Scholar, 7 papers were published on ATRP in 1995, 47 in 1996, 111 in 1997, 150 in 1998, 318 in 1999, and more than 300 in 2000. In addition, many papers using the ATRP concept but not using the ATRP name are being published (alternative nomen- clature include transition metal mediated living radi- cal polymerization, transition metal catalyzed living free-radical polymerization, atom transfer polymer- ization, etc.). II. Mechanistic Understandings of Atom Transfer Radical Polymerization A. Components As a multicomponent system, ATRP is composed of the monomer, an initiator with a transferable (pseudo)halogen, and a catalyst (composed of a transition metal species with any suitable ligand). Sometimes an additive is used. For a successful ATRP, other factors, such as solvent and tempera- ture, must also be taken into consideration. 1. Monomers A variety of monomers have been successfully polymerized using ATRP. Typical monomers include styrenes, (meth)acrylates, (meth)acrylamides, and acrylonitrile, which contain substituents that can stabilize the propagating radicals.22,23 Ring-opening polymerization has been also successful.26,27 Even under the same conditions using the same catalyst, each monomer has its own unique atom transfer equilibrium constant for its active and dormant species. In the absence of any side reactions other than radical termination by coupling or dispropor- tionation, the magnitude of the equilibrium constant (Keq ) kact/kdeact) determines the polymerization rate. ATRP will not occur or occur very slowly if the equilibrium constant is too small. In contrast, too large an equilibrium constant will lead to a large amount of termination because of a high radical concentration. This will be accompanied by a large amount of deactivating higher oxidation state metal complex which will shift the equilibrium toward dormant species and may result in the apparently slower polymerization.28 Each monomer possesses its own intrinsic radical propagation rate. Thus, for a specific monomer, the concentration of propagating radicals and the rate of radical deactivation need to be adjusted to maintain polymerization control. However, since ATRP is a catalytic process, the overall position of the equilibrium not only depends Scheme 2. Transition-Metal-Catalyzed ATRP Atom Transfer Radical Polymerization Chemical Reviews, 2001, Vol. 101, No. 9 2923

on the radical (monomer) and the dormant species, but also can be adjusted by the amount and reactivity of the transition-metal catalyst added (cf. eq 2) 2. Initiators The main role of the initiator is to determine the number of growing polymer chains. If initiation is fast and transfer and termination negligible, then the number of growing chains is constant and equal to the initial initiator concentration. The theoretical molecular weight or degree of polymerization (DP) increases reciprocally with the initial concentration of initiator in a living polymerization (eq 1). Figure 1 illustrates a linear increase of molecular weights with conversion. Simultaneously, polydis- persities (Mw/Mn) decrease with the conversion, de- pending on the relative rate of deactivation (cf. eq 3). In ATRP, alkyl halides (RX) are typically used as the initiator and the rate of the polymerization is first order with respect to the concentration of RX. To obtain well-defined polymers with narrow molecular weight distributions, the halide group, X, must rapidly and selectively migrate between the growing chain and the transition-metal complex. Thus far, when X is either bromine or chlorine, the molecular weight control is the best. Iodine works well for acrylate polymerizations in copper-mediated ATRP29 and has been found to lead to controlled polymeri- zation of styrene in ruthenium- and rhenium-based ATRP.30,31 Fluorine is not used because the C-F bond is too strong to undergo homolytic cleavage. Some pseudohalogens, specifically thiocyanates and thio- carbamates, have been used successfully in the polymerization of acrylates and styrenes.29,32,33 Initiation should be fast and quantitative with a good initiator. In general, any alkyl halide with activating substituents on the R-carbon, such as aryl, carbonyl, or allyl groups, can potentially be used as ATRP initiators. Polyhalogenated compounds (e.g., CCl4 and CHCI3) and compounds with a weak R-X bond, such as N-X, S-X, and O-X, can also be used as ATRP initiators. When the initiating moiety is attached to macromolecular species, macroinitiators are formed and can be used to synthesize block/graft copolymers.21 Similarly, the efficiency of block/graft copolymerization may be low if the apparent rate constant of cross-propagation is smaller than that of the subsequent homopolymerization. It should be noted, however, that R-X bonds can be cleaved not only homolytically but also heterolyti- cally, which depends mostly on the initiator structure and the choice of the transition metal catalyst. For example, side reactions are observed for copper- mediated ATRP of p-methoxystyrene, likely due to the heterolytic cleavage of C-X bond or oxidation of the radical to the corresponding carbocation.14,34 3. Catalysts Perhaps the most important component of ATRP is the catalyst. It is the key to ATRP since it determines the position of the atom transfer equi- librium and the dynamics of exchange between the dormant and active species. There are several pre- requisites for an efficient transition metal catalyst. First, the metal center must have at least two readily accessible oxidation states separated by one electron. Second, the metal center should have reasonable affinity toward a halogen. Third, the coordination sphere around the metal should be expandable upon oxidation to selectively accommodate a (pseudo)- halogen. Fourth, the ligand should complex the metal relatively strongly. Eventually, the position and dynamics of the ATRP equilibrium should be ap- propriate for the particular system. A variety of transition-metal complexes have been studied as ATRP catalysts and will be discussed in more detail later in this paper. 4. Solvents ATRP can be carried out either in bulk, in solution, or in a heterogeneous system (e.g., emulsion, suspen- sion). Various solvents, such as benzene, toluene, anisole, diphenyl ether, ethyl acetate, acetone, dim- ethyl formamide (DMF), ethylene carbonate, alcohol, water, carbon dioxide, and many others, have been used for different monomers. A solvent is sometimes necessary, especially when the obtained polymer is insoluble in its monomer (e.g., polyacrylonitrile). Several factors affect the solvent choice. Chain transfer to solvent should be minimal. In addition, interactions between solvent and the catalytic system should be considered. Catalyst poisoning by the solvent (e.g., carboxylic acids or phosphine in copper- based ATRP)35 and solvent-assisted side reactions, such as elimination of HX from polystyryl halides, which is more pronounced in a polar solvent,36 should be minimized. The possibility that the structure of the catalyst may change in different solvents should also be taken into consideration. For example, the ATRP of n-butyl acrylate with CuBr(bpy)3 (bpy ) 2,2���-bipyridine here and below the notation of the complex reflects only the stoichiometry of added reagents and NOT the structure of the complex) as the catalyst carried out in ethylene carbonate was found to proceed much faster than in bulk.37 A structural change from a dimeric halogen-bridged Cu(I) species in the bulk system to a monomeric Cu(I) species in ethylene carbonate was proposed to explain the rate difference. A similar rate enhancement in polar media was observed later from different studies.38-40 Polar media can also help to dissolve the catalyst. For example, homogeneous ATRP using CuBr(bpy)3 was achieved using 10% v/v DMF.41 Figure 1. Schematic representation of the evolution of the molecular weights and polydispersities with conversion for a living polymerization. DP ) [M]0/[initiator]0 �� conversion (1) 2924 Chemical Reviews, 2001, Vol. 101, No. 9 Matyjaszewski and Xia

5. Temperature and Reaction Time The rate of polymerization in ATRP increases with increasing temperature due to the increase of both the radical propagation rate constant and the atom transfer equilibrium constant. As a result of the higher activation energy for the radical propagation than for the radical termination, higher kp/kt ratios and better control (���livingness���) may be observed at higher temperatures. However, chain transfer and other side reactions become more pronounced at elevated temperatures.36,42 In general, the solubility of the catalyst increases at higher temperatures however, catalyst decomposition may also occur with the temperature increase.43,44 The optimal tempera- ture depends mostly on the monomer, the catalyst, and the targeted molecular weight. At high monomer conversions, the rate of propaga- tion slows down considerably however, the rate of any side reaction does not change significantly, as most of them are monomer concentration indepen- dent. Prolonged reaction times leading to nearly complete monomer conversion may not increase the polydispersity of the final polymer but will induce loss of end groups.45 Thus, to obtain polymers with high end-group functionality or to subsequently synthesize block copolymers, conversion must not exceed 95% to avoid end-group loss. 6. Additives Additives are sometimes essential for a successful ATRP. For example, a Lewis acid, such as aluminum and other metal alkoxides, is needed for the con- trolled polymerization of MMA catalyzed by RuCl2- (PPh3)3 or other systems.15,46,47 No or very slow poly- merization was observed in the absence of the Lewis acid activator. Presumably, the aluminum compound can activate and stabilize the catalyst in the higher oxidation state.46 Polymerization in the presence of very polar solvents such as water can be acceler- ated.39 The presence of strong nucleophiles such as phosphines may sometimes terminate the process. 35 B. Typical Phenomenology 1. Kinetics The kinetics of ATRP is discussed here using copper-mediated ATRP as an example. Mechanistic investigations into ATRP based upon other metal systems are anticipated to yield similar results. According to Scheme 2 using the assumption that contribution of termination becomes insignificant due to the persistent radical effect7,48 (PRE) (especially for the chain-length-dependent PRE49) and using a fast equilibrium approximation, which is necessary for observed low polydispersities, the rate law (eq 2, cf. Scheme 1 for the explanation of all symbols) for ATRP can be derived as follows. Figure 2 shows a typical linear variation of conver- sion with time in semilogarithmic coordinates. Such a behavior indicates that there is a constant concen- tration of active species in the polymerization and first-order kinetics with respect to monomer. How- ever, since termination occurs continuously, the concentration of the Cu(II) species increases and deviation from linearity may be observed. For the ideal case with chain length independent termina- tion, PRE kinetics implies the semilogarithmic plot of monomer conversion vs time to the 2/3 exponent should be linear.7 Nevertheless, a linear semiloga- rithmic plot is often observed. This may be due to an excess of the Cu(II) species present initially, a chain- length-dependent termination rate coefficient, and heterogeneity of the reaction system due to limited solubility of the copper complexes. It is also possible that self-initiation may continuously produce radicals and compensate for termination.50,51 Similarly, ex- ternal orders with respect to initiator and the Cu(I) species may also be affected by the PRE.52 Results from kinetic studies of ATRP for styrene,35 methyl acrylate (MA),53 and methyl methacrylate (MMA)54,55 under homogeneous conditions indicate that the rate of polymerization is first order with respect to monomer, initiator, and Cu(l) complex concentrations. These observations are all consistent with the derived rate law (eq 2). The kinetically optimum ratio of ligand to copper in the polymeri- zation of both styrene and MA was determined to be 2:1. Below this ratio the polymerization rate was usually slower, and above this ratio the polymeriza- tion rate remained constant. It should be noted that the optimum ratio can vary with regard to changes in the monomer, counterion, ligand, temperature, and other factors.43,54,56 The precise kinetic law for the deactivator (X-CuII) was more complex due to the spontaneous generation of Cu(II) via the persistent radical effect.7,35,52 In the atom transfer step, a reactive organic radical is generated along with a stable Cu(II) species that can be regarded as a persistent metalloradical. If the initial concentration of deactivator Cu(II) in the polymerization is not sufficiently large to ensure a fast rate of deactivation (kdeact[Cu(II)]), then coupling of the organic radicals will occur, leading to an increase in the Cu(II) concentration. This process has been observed experimentally using IH NMR, UV- vis, EPR, and GC-MS techniques.35,57 With each radical termination event, 2 equiv of Cu(II) will form irreversibly. Radical termination occurs rapidly until a sufficient amount of deactivator Cu(II) is formed and the radical concentration becomes low enough. Under such conditions, the rate at which radicals combine (kt[Rl]2) will become much slower than the Rp ) kp[M][P*] ) kpKeq[M][I]0 �� [CuI]/[X - CuII] (2) Figure 2. Schematic representation of the dependence of the conversion on time in linear and semilogarithmic coordinates. Atom Transfer Radical Polymerization Chemical Reviews, 2001, Vol. 101, No. 9 2925

rate at which radicals react with the copper(ll) complex (kdeact[Rl][Cu(II)]) in a deactivation process and a controlled/���living��� polymerization will proceed. Typically, a small fraction (���5%) of the total growing polymer chains will be terminated during the early stage of the polymerization, but the majority of the chains ( 90%) will continue to grow successfully. If a small amount of the deactivator (���10 mol %) is added initially to the polymerization, then the pro- portion of terminated chains can be greatly re- duced.20,57 The effect of Cu(II) on the polymerization may additionally be complicated by its poor solubility, by a slow reduction by reaction with monomers leading to 1,2-dihaloadducts, or from the self-initiated systems such as styrene and other monomers.51,58 2. Molecular Weight Similarly to a typical living polymerization, the average molecular weight of the polymer made by a well-controlled ATRP can be predetermined by the ratio of consumed monomer and the initiator (DPn ) ���[M]/[I]o. DP ) degree of polymerization) while maintaining a relatively narrow molecular weight distribution (1.0 Mw/Mn 1.5). In addition, precise control over the chemistry and the structure of the initiator and active end group allows for the synthesis of end-functionalized polymers and block copolymers. Well-defined polymers with molecular weights rang- ing from 1000 to 150 000 have been successfully synthesized. However, termination and other side reactions are also present in ATRP, and they become more prominent as higher molecular weight polymers are targeted. For example, in the copper-mediated ATRP of styrene, a slow termination process was observed arising mainly from the interaction of the copper(II) species with both the growing radical and the macromolecular alkyl halide. This effect is neg- ligible for low molecular weight polystyrene but could result in an upper limit to styrene ATRP.36 Figure 3 shows a typical linear increase of the molecular weights with conversion in the ATRP of methyl acrylate.53 Since the rate constants of propa- gation for acrylates are relatively large, initially, higher polydispersities were observed because several monomer units are added during each activation step. However, with the progress of the reaction, chains become more uniform due to continuous exchange reactions. The polydispersities drop with conversion, as predicted by eq 3. If kp and the concentrations of initiator and deactivator are known, the rate constant of deactivation can be calculated from the evolution of polydispersities with conversion. 3. Molecular Weight Distribution The molecular weight distribution or polydispersity (Mw/Mn) is the index of the polymer chain-length distribution. In a well-controlled polymerization, Mw/ Mn is usually less than 1.10. Equation 3 illustrates how the polydispersity index in ATRP in the absence of significant chain termination and transfer relates to the concentrations of initiator (RX) and deactivator (D), the rate constants of propagation (kp) and deactivation (kdeact), and the monomer conversion (p).59 This equation holds for conditions when initia- tor is completely consumed and degrees of polymer- ization are sufficiently high otherwise the Poisson term should be added (1/DPn). Thus, for the same monomer, a catalyst that deactivates the growing chains faster will result in polymers with lower polydispersities (smaller kp/ kdeact). Alternatively, polydispersities should decrease with an increasing concentration of deactivator, although at the cost of slower polymerization rates. For example, the addition of a small amount of Cu- (II) halides in the copper-based ATRP leads to better controlled polymerizations with decreased polymer- ization rates.53,60 Perhaps most important, however, is the propagation rate constant higher polydisper- sities are usually found for polyacrylates than for polystyrene or polymethacrylates due to a much higher kp for the former monomers.61 Other predic- tions from eq 3 include higher polydispersities for shorter chains (higher [RX]o) and a decrease of the polydispersity with increasing monomer conversion. The implications of eq 3 are in agreement with the experimental results. It is also possible to correlate Figure 3. Evolution of molecular weight and polydispersity in the ATRP of MA: T ) 90 ��C [MA]o ) 11.2 M [MA]o/ [MBP]o ) 1513 (MBP ) methyl 2-bromopropionate) [MBP]o/[CuBr]o/[dTbpy]o ) 1/1/2 (dTbpy ) 4,4���-di-tert-butyl-2,2���- bypyridine). Mw/Mn ) 1 + (kdeact[Dp [RX]0k ] 29( 2 p - 1 (3) 2926 Chemical Reviews, 2001, Vol. 101, No. 9 Matyjaszewski and Xia

polydispersities with the rate constant of activation when they are plotted against time rather than conversion.62 The rate constant of deactivation (kdeact) is affected by a number of factors, such as the transition metal, the metal counterion, and the ligand. For the same catalytic system, an important factor is the lability of the X-Mt bond in the deactivator. CuBr2(dNbpy)2 (dNbpy ) 4,4���-di(5-nonyl)-2,2���-bipyridine) yields faster deactivation than CuCl2(dNbpy)2. Similar results were obtained in earlier studies on the efficiency of inhibition of various metal salts.13 In ATRP, the concentration of deactivator increases sharply at the beginning of the polymerization and then increases slowly, but continuously, with mono- mer conversion.63 The addition of a small amount of Cu(II) halides at the beginning of the polymerization can reduce the proportion of terminated chains and help establish the atom transfer equilibrium. Con- versely, the addition of small amounts of copper(0) in copper-mediated ATRP can result in a faster polymerization rate, as ���excess��� copper(II) is reduced to copper(I) (cf. section II.G).64 It should be noted that deactivators may also participate in side reactions. For example, reduced molecular weights and termination were observed when the ATRP of styrene was carried out in the presence of a large amount of cupric triflate, likely due to the oxidation of growing radicals via an outer- sphere electron-transfer process.65 Similarly, cuprous triflate and bromide may reduce the growing radicals in the polymerization of acrylonitrile66 and methyl acrylate.65 4. Normal and Reverse ATRP In a normal ATRP, the initiating radicals are generated from an alkyl halide in the presence of a transition metal in its lower oxidation state (e.g., CuBr(dNbpy)2) however, conventional radical initia- tors can also be employed. For example, ATRP can be initiated using azobisisobutyronitrile (AIBN) with the transition-metal compound in its higher oxidation state (e.g., CuBr2(dNbpy)2). The latter approach has been named reverse ATRP and was successfully used for copper-based heterogeneous67-70 and homoge- neous71 systems in solution and in emulsion72 as well as for iron complexes.73 Other conventional radical initiators have also been used for reverse ATRP. For example, 1,1,2,2-tetraphenyl-1,2-ethanediol (TPED)74 and diethyl 2,3-dicyano-2,3-diphenylsucci- nate (DCDPS)75 have been used successfully in the presence of FeCl3(PPh3)3 for the reverse ATRP of MMA and styrene, respectively. For TPED, PMMA with Mn ) 171 800 and Mw/Mn ) 1.13 was obtained but the initiation efficiency was low (0.5). For DCDPS, the experimental molecular weights by size exclusion chromatography (SEC) were lower than the calcu- lated values, assuming one molecule of DCDPS generated two living polymer chains. More recently, the reverse ATRP using tetraethylthiuram disulfide and FeCl3(PPh3)3 as the initiating system resulted in the formation of PMMA with Mn ��� 7000 and Mw/ Mn ) 1.05 within 8 min at 90 ��C in bulk.76 The reverse ATRP initiated by peroxides some- times behaves quite differently than that based on the azo compounds. For instance, no control over the polymerization was observed for the homogeneous BPO/CuBr2(dNbpy)2 system (BPO ) benzoyl perox- ide). In contrast, controlled/���living��� polymerization was observed when BPO was used together with CuBr(dNbpy)2. The differences between the BPO and AIBN systems are ascribed to an electron transfer and the formation of a copper benzoate species.70 In a heterogeneous system using bpy as the ligand, both CuBr and CuBr2 yielded a controlled polymerization of styrene.69 5. Experimental Setup ATRP can be carried out either in bulk or with a solvent. Solvents are often used to alleviate viscosity problems that arise at high conversions. As discussed previously, a variety of solvents can be used in ATRP. Environmentally friendly media, such as water72,77-81 and carbon dioxide,82 have been used. Depending on the initial conditions, ATRP can be performed in solution, suspension,79,83 emulsion,72,77,84 miniemul- sion,85 or dispersion.82 Kinetics of ATRP in emulsion is quite different from conventional emulsion polymerization.86 Due to the slow growth of MW with conversion, the mech- anism of nucleation changes entirely. Moreover, partition coefficients of both activators and deactiva- tors in organic and aqueous phases become very important. The catalytic system should preferentially reside in the organic phase but should also be slightly soluble in water to transfer between monomer drop- lets and growing particles and also to scavenge radicals in water.86 Both normal and reverse ATRP has been successful, although colloidal stability of latexes is higher and particle size smaller for the reverse ATRP.86 The concept of compartmentaliza- tion, which is the essence of emulsion polymerization, is strongly related to the living polymerization. The proportion of terminated chains can be smaller than in bulk at the same overall rate of monomer con- sumption. However, only when the size of growing particles is smaller than 50 nm does the effect become significant.87 6. Catalyst Homogeneity Both heterogeneous and homogeneous catalytic systems have been used in ATRP. Better solubility of the transition-metal complex is achieved by adding long alkyl substituents to the ligand.35,43,88 Homoge- Scheme 3. Reverse ATRP Using AIBN as the Initiator Atom Transfer Radical Polymerization Chemical Reviews, 2001, Vol. 101, No. 9 2927

neous systems allow for the detailed kinetic and mechanistic studies of the polymerization.35,55,88 In addition, polymers with lower polydispersities are usually obtained with a homogeneous catalyst due to a higher concentration of deactivator in solution.89 Attempts have also been made using solid sup- ported catalysts.90-94 The usual procedure for remov- ing the catalyst from a reaction on a laboratory scale involves precipitating the polymer or filtering the polymer solution through a column of aluminum oxide, which adsorbs the catalyst. Removal of the copper-based catalyst using an ion-exchange resin has also been reported.95 The disadvantages of these techniques include cost, problems with scale-up, loss of polymer, and difficulties in separating the catalyst from functional polymers that interact with the copper complexes. Immobilization of the catalytic system on a solid support provides a more efficient way of separating and potentially recycling the catalyst. Thus, multidentate nitrogen donor ligands as well as Schiff base ligands have been covalently bounded to silica and cross linked polystyrene sup- ports. In general, polymers with higher polydisper- sities (Mw/Mn 1.5) were obtained using the solid supported catalysts. This was explained by slow deactivation of the growing radicals resulting from slow diffusion toward the metal center. Lower poly- dispersities were obtained when the catalyst was physically absorbed onto a solid support however, only the controlled polymerization of methacrylates have been reported so far.91,93,94 Other approaches involve the reversible adsorption of the transition- metal complex using ion-exchange resins,95 a hybrid catalyst system consisting of majority of the im- mobilized catalyst and a minute amount of soluble more active catalyst,96 or using ligands whose solu- bility is strongly dependent on the temperature.97 7. Summary and Outlook The current understanding of the kinetics and mechanism of ATRP allows for a basic correlation of the effect of concentrations and structures of the involved reagents on the polymerization rates, mo- lecular weights, and polydispersities. The structural effects will be discussed in more detail in the subse- quent sections. ATRP is more complex than other CRP methods because it involves a complex, often heterogeneous catalytic system. The solubility, struc- ture, concentration in solution, aggregation, effect of ion pairing, etc., may change not only with the overall catalyst composition and preparation method but also for each monomer, solvent, and temperature. Thus, more detailed information on the structure of both activator and deactivator in solution is needed. Additional complications appear in aqueous systems, both homogeneous and heterogeneous. In aqueous solution halogens can be displaced from transition metals (hydrolyze) and significantly reduce the con- centration of the true deactivator (X-Mtn+1 species). Complexes may be strongly solvated by water, reduc- ing rates of activation and deactivation. Ligands may be more labile, enabling reorganization of the cata- lytic system. In heterogeneous systems, especially emulsion, the behavior of ATRP and other CRPs is very different than conventional RP. Due to the slow growth of MW with conversion, the mechanism of nucleation changes entirely. Moreover, partition coefficients of both activator and deactivator in organic and aqueous phases become very important. Preferentially, the catalytic system should reside in the organic phase but should be also slightly soluble in water to scavenge radicals and transfer between monomer droplets and growing particles. The concept of com- partmentalization, which is the essence of emulsion polymerization, has a strong effect in the living polymerization. It can potentially reduce the propor- tion of terminated chains however, only when size of growing particles is smaller than 50 nm does the effect become significant, i.e., proportion of termi- nated chains becomes lower than in bulk at the same overall rate of monomer consumption. Perhaps one of the main challenges for the com- mercialization of the ATRP process is the removal and recycling of the catalyst. There are several approaches being actively evaluated which are based on immobilization, biphasic systems with water, ionic liquids, and fluorinated solvents. More efficient meth- ods of removal by extraction, filtration, etc., are needed. Another approach is to continuously increase the activity of the catalytic system, which may enable reducing the amount of the catalyst to a level that it may be left in the final polymer. Nearly all ATRP reactions are carried out in batch or semibatch systems, and conversion to continuous systems should be studied, perhaps using bulk monomer but reach- ing only partial monomer conversion in each cycle. C. ATRP Monomers Various monomers have been successfully polym- erized using ATRP: styrenes, (meth)acrylates, (meth)- acrylamides, dienes, acrylonitrile, and other mono- mers which contain substituents that can stabilize the propagating radicals. Ring-opening polymeriza- tion is also possible. However, even using the same catalyst under the same conditions, each monomer has its own unique atom transfer equilibrium con- stant for its active and dormant species. The product of kp and the equilibrium constant (Keq ) kact/kdeact) essentially determines the polymerization rate. ATRP will occur very slowly if the equilibrium constant is too small. This is plausibly the main reason why polymerization of less reactive monomers such as olefins, halogenated alkenes, and vinyl acetate has not yet been successful. Because each monomer has a specific equilibrium constant, optimal conditions for polymerization which include concentration and type of the catalyst, temperature, solvent, and some additives may be quite different. Therefore, we discuss ATRP monomers separating them into dif- ferent groups starting from non-polar styrenes, fol- lowed by various (meth)acrylate esters, nitriles, amides, acids, and other monomers. 1. Styrenes ATRP of styrene and its derivatives has been reported for the copper,17-19,35 iron,98 ruthenium,31 and rhenium30 catalytic systems thus far the major- ity of the work has been performed using the copper- based systems. 2928 Chemical Reviews, 2001, Vol. 101, No. 9 Matyjaszewski and Xia

In addition to 1-phenylethyl halide and benzylic halides, a variety of compounds, such as allylic halides and functional R-haloesters,99 polyhaloge- nated alkanes,18,100 and arenesulfonyl chlorides,55 have been used successfully as the initiators for the copper-mediated styrene ATRP. One of the most extensively studied systems is the polymerization of styrene conducted at 110 ��C with CuBr(dNbpy)2 as the catalyst and alkyl bromides as the initiators. A similar system for the chloride-mediated polymeri- zation is conducted at 130 ��C to obtain similar poly- merization rates.35 The reaction temperature can be lowered to 80-90 ��C to produce well-defined poly- styrenes in a reasonable time with the use of a more efficient catalyst, such as CuBr/PMDETA (PMDETA ) N,N,N���,N������,N������-pentamethyldiethylenetriamine)101 or CuOAc/CuBr/dNbpy.60 However, to maintain a sufficiently large propagation rate, avoid vitrification at high conversion (for polystyrene Tg ��� 100 ��C), and sometimes increase the solubility of the catalysts, higher reaction temperatures ( 100 ��C) are preferred for styrene ATRP. The reaction may be carried out in bulk or using a solvent, but the stability of the halide end group displays a pronounced solvent dependence as demonstrated by model studies using 1-phenylethyl bromide. As a result, nonpolar solvents are recommended for styrene ATRP.36 Polystyrenes with molecular weights (Mn) ranging from 1000 to 100 000 with low polydispersities have been prepared. Better molecular weight control is obtained at lower temperatures, presumably due to a lower contribution of the thermal self-initiation.42,58 Additionally, a wide range of styrene derivatives with different substituents on the aromatic ring have been polymerized in a well-controlled fashion.34 Well-de- fined p-acetoxystyrene was prepared, and subsequent hydrolysis afforded water-soluble poly(vinylphe- nol).102 In general, styrenes with electron-withdraw- ing substituents polymerize faster. The Hammett correlation for ATRP of styrene provided F ) 1.5 com- pared to F ) 0.5 for the radical propagation constants. This indicates that the atom transfer equilibrium was more shifted toward the active species side for sty- renic monomers bearing electron-withdrawing groups. This behavior was explained by the higher ATRP reactivity of secondary benzylic halides with electron- withdrawing groups.103 Scheme 4 shows some styrene derivatives successfully polymerized by ATRP. 2. Acrylates The controlled ATRP of acrylates has been reported for copper-,16,18,53 ruthenium-,104 and iron-based sys- tems.105 Copper appears to be superior over other transition metals in producing well-defined poly- acrylates with low polydispersities in a relatively short time. This is partially due to the fast deactiva- tion of the growing acrylic radicals by the cupric halides. Typically polymerizations were conducted in bulk with an alkyl 2-bromopropionate initiator. Well- defined polyacrylates with Mn up to 100 000 and Mw/ Mn 1.1 were prepared. Depending on the catalyst, a wide range of polymerization temperatures are possible to produce polymers within a reasonable time (e.g., Mn ) 20 000 in ca. 2 h). For example, using 0.05 mol % of CuBr/Me6TREN (Me6TREN ) tris[2- (dimethylamino)ethyl]amine) as the catalyst, poly- (MA) with Mn ) 12 600 and Mw/Mn ) 1.10 was obtained in 1 h at ambient temperature.106 A wide range of acrylates with various side chains have been polymerized using ATRP (Scheme 5). For example, well-defined functional polymers were ob- tained by the ATRP of 2-hydroxyethyl acrylate (HEA)80,107 and glycidyl acrylate.108 Poly(tert-butyl acrylate) was also prepared in a well-controlled fashion.109 Subsequent hydrolysis yields well-defined poly(acrylic acid). In addition, well-defined homopoly- mer and block copolymers with long alkyl chain142 and fluorocarbon side chains have been prepared.82,110 When allyl acrylate was subjected to ATRP condi- tions with bpy or dNbpy as the ligand, a cross-linking reaction occurred, even at 0 ��C.58 3. Methacrylates ATRP of methyl methacrylate (MMA) has been reported for ruthenium,15,104 copper,111,112 nickel,113-115 iron,98,116,117 palladium,118 and rhodium119 catalytic systems. The facile polymerizability of MMA and the large range of available catalysts for the ATRP reaction is due to the relative ease of activation of the dormant species and the high values of the ATRP equilibrium constants. The equilibrium constants can sometimes be too high to obtain a controlled ATRP process, as is the case for the Me6TREN ligands.28 Using the known rate constant of propagation for Scheme 4. Various Styrenes Polymerized by ATRP Scheme 5. Representative Acrylates Polymerized by ATRP Atom Transfer Radical Polymerization Chemical Reviews, 2001, Vol. 101, No. 9 2929

MMA, typical radical concentrations for the bulk and solution controlled ATRP of MMA are estimated to be between 10-7 and l0-9 M. Most polymerizations of MMA were carried out in solution at temperatures ranging from 70 to 90 ��C. Solvents are necessary to solubilize the forming poly- (MMA) (PMMA), which has a glass transition tem- perature Tg 100 ��C. In addition, solution polym- erization helps to keep the concentration of growing radicals low. Under comparable conditions, the cop- per-mediated ATRP of MMA displays a significantly higher equilibrium constant when compared with styrene and MA. As a result, higher dilution and a lower catalyst concentration should be used for the MMA polymerization. Initiation plays an important role in the ATRP of MMA. The best initiators include sulfonyl chlorides111 and 2-halopropionitrile98 because these initiators have sufficiently large apparent rate constants of initiation (high atom transfer equilibrium constants). Well-defined PMMA can be prepared within the molecular weight range from 1000 to 180 000. A series of initiators, including chloromethanes, R-chlo- roesters, R-chloroketones, and R-bromoesters, were studied in ruthenium-mediated ATRP of MMA.120 CCl3COCH3, CHCl2COPh, and dimethyl 2-bromo- 2,4,4-trimethylglutarates were among the best initia- tors, yielding PMMA with controlled molecular weights and low polydispersities (Mw/Mn ) 1.1-1.2). Similar studies were performed for Cu-based sys- tems.121,122 It should be noted that some of these initiators are too active for the copper-based systems and lead to excessive termination or other side reactions.123 Other methacrylic esters have also been success- fully polymerized. These include n-butyl methacry- late,55,77,88,124 2-(dimethylamino)ethyl methacrylate (DMAEMA),125 2-hydroxyethyl methacrylate (HE- MA)104,126 and silyl-protected HEMA,127 methacrylic acid in its alkyl protected form128 or as its sodium salt,129 methacrylates with an oligo(ethylene oxide) substituent,39 and fluorinated methacrylic es- ters.82,110,130 Scheme 6 illustrates examples of meth- acrylates polymerized by ATRP. 4. Acrylonitrile Metal mediated controlled radical polymerization of acrylonitrile has so far only been reported for copper-mediated ATRP.66,131,132 It is necessary to use a solvent because polyacrylonitrile is not soluble in its monomer. DMF is a good solvent for polyacryloni- trile however, it may also complex with copper and deactivate the catalyst. Successful polymerizations have been carried out in ethylene carbonate in the presence of the CuBr(bpy)2 complex using R-bro- mopropionitrile as the initiator at temperatures from 44 to 64 ��C. The CuBr(bpy)2 catalyst was soluble in the strongly polar polymerization medium, and the system was homogeneous. Well-defined polyacryloni- trile with Mw/Mn 1.05 has been prepared within the molecular weight range from 1000 to 10 000. In all polymerizations there was significant curvature in the first-order kinetic plot of the monomer con- sumption. 1H NMR spectroscopy and MALDI-TOF analysis showed that some halide end groups were irreversibly removed during the polymerization. It was proposed that the reduction of the propagating radical by the cuprous halide to form an anion was the major chain termination reaction.66 Acrylonitrile has also been copolymerized with styrene in a well- controlled fashion to yield gradient copolymers with molecular weights ranging from 1000 to 15 000.133 5. (Meth)acrylamides Polymers of acrylamide and its derivatives have found wide use in industry, agriculture, and medicine owing to their remarkable properties such as water solubility and potential biocompatibility. There are a few reports on the attempted ATRP of acrylamide. Using CuCl-bpy as the catalyst and surface-bound benzyl chloride as the initiator, Wirth et al. made poly(acrylamide) films from a silica surface.134 The resulting materials provided good analytical separa- tions however, detailed proof for the controlled character of the polymerization was not provided. Li and Brittain also attempted the controlled polymer- ization of acrylamide by ATRP but did not obtain any polymers using CuBr(bpy)3 as the catalyst and 1-(bro- moethyl)benzene as the initiator at various temper- atures.135 It was shown using model compounds and kinetic studies that the polymerization of acrylamide under typical ATRP conditions displayed a much lower ATRP equilibrium constant than the acrylates or styrene.136 The inactivation of the catalyst by complexation of copper by the forming polymer and displacement of the terminal halogen atom by the amide group are two potential side reactions. Inter- estingly, using 1,4,8,11-tetramethyl-1,4,8,11-tetraaza- cyclotetradecane (Me4Cyclam) as a ligand provided polymers in high yields in a short time. Unfortu- nately, the polymerization was not controlled and displayed slow deactivation characteristics. Loss of the chain-end halogen was considered previously137 and recently confirmed by end-group analysis through the use of mass spectrometry.138 The conclusion is that the presence of the metal as a Lewis acid in ATRP and its complexation to the amide functionality slows deactivation and makes the process an uncon- trolled polymerization. Nevertheless, by using the Me4Cyclam-based catalytic system and well-defined macroinitiators prepared by ATRP, block copolymers Scheme 6. Various Methacrylates Polymerized by ATRP 2930 Chemical Reviews, 2001, Vol. 101, No. 9 Matyjaszewski and Xia

of poly(methyl acrylate-b-N,N-dimethylacrylamide) (Mn ) 4800, Mw/Mn ) 1.33) and poly(n-butyl acrylate- b-N-(2-hydroxypropyl)methacrylamide] (Mn ) 34 000, Mw/Mn ) 1.69) were synthesized.136 The best results for the ATRP of (meth)acrylamide were obtained using one of the most powerful catalytic systems (CuCl/Me6TREN) due to its high equilibrium con- stant. Moreover, polymerizations were carried out using alkyl chlorides as the initiators at low temper- ature (20 ��C) in a low polarity solvent (toluene) to minimize side reactions.139 For example, poly(N,N- dimethylacrylamide) with molecular weight Mn ) 8400 and polydispersity Mw/Mn ) 1.12 was formed at room temperature in 50% toluene solution. Metals other than copper have also been studied in the ATRP of acrylamide. For example, living polymerization of dimethylacrylamide (DMAA) is possible with the use of a bromide initiator such as CCl3Br in conjunction with RuCl2(PPh3)3 and Al(Oi- Pr)3 in toluene at 60 ��C.140 Polymers with relatively high polydispersities (Mw/Mn ) 1.6) were obtained. Better control was achieved at lower temperatures, presumably due to a lower contribution of the side reactions. A unique amide monomer, N-(2-hydroxypropyl) methacrylamide, was polymerized in a controlled manner using CuBr/Me4Cyclam as the catalyst.141 The polymerization was carried out in 1-butanol to yield a relatively well-defined polymer (Mn ) 21 300, Mw/Mn ) 1.38) and block copolymers. 6. (Meth)acrylic Acids Controlled polymerization of (meth)acrylic acid by ATRP presents a challenging problem because the acid monomers can poison the catalysts by coordinat- ing to the transition metal. In addition, nitrogen- containing ligands can be protonated, which inter- feres with the metal complexation ability. Recently, Armes and co-workers reported the successful ATRP of sodium methacrylate in water using CuBr(bpy)3 as the catalyst with a poly(ethylene oxide)-based macroinitiator.129 Yields were moderate to good, molecular weight control was good, and the polydis- persities were low (Mw/Mn ) 1.30) however, high polydispersities were observed when the target Mn 10 000. The choice of pH and initiator was critical. The optimum pH lies between 8 and 9, as there appears to be a balance between the reduced propa- gation rate at high pH and competing protonation of the ligand at low pH. In addition, low conversion and initiator efficiency were obtained when sodium 2-bro- moisobutyrate was used as the initiator. Other acidic monomers such as sodium vinylbenzoate were also successfully polymerized in aqueous media using a similar methodology.143 Alternatively, poly(meth)acrylic acids can be pre- pared by polymerization of protected monomers such as trimethylsilyl methacrylate, tert-butyl methacry- late, tetrahydropyranyl methacrylate, and benzyl methacrylate.144 7. Miscellaneous Monomers Pyridine-containing polymers are useful for various applications such as water-soluble polymers and coordination reagents for transition metals. Both 4-vinylpyridine (4VP) and poly(4-vinylpyridine) (P4VP) can act as coordinating ligands for transition metals and compete for the binding of the metal catalysts in ATRP. By employing a strongly coordinating ligand such as Me6TREN, well-defined P4VP has been obtained at 40 ��C using a copper-based catalytic system.145 Alternating copolymers of isobutene with MA, BA, and AN have been prepared using CuBr(bpy)3 as the catalyst and 1-phenylethyl bromide as the initiator at 50 ��C.146 The experimental molecular weights were close to the theoretical values, ranging from 4000 to 50 000. The polydispersities were relatively high (Mw/ Mn ��� 1.50). Evidence of the alternating sequences and the tacticity of the isobutene with the MA was provided by 1H NMR analysis. The prepared alter- nating copolymer with MA was an elastomer with a preponderant syndiotactic structure and a low glass transition temperature (Tg ��� -30 ��C). Alternating copolymerizations of maleimides with styrene22,146-149 and MMA150 have been carried out using copper-based ATRP. A linear increase of Mn with conversion was observed up to Mn ��� 13 000, with Mw/Mn around 1.16-1.36. N-(2-acetoxyethyl)- maleimide was found to copolymerize faster than N-phenylmaleimide.147 Polymerization of vinylidene chloride and isoprene 17 by copper-mediated ATRP has also been carried out. Controlled polymerization of vinyl acetate (VOAc) by ATRP remains challenging, largely due to the small atom transfer equilibrium constant.151 How- ever, the successful copolymerization of VOAc with MA has been reported.99 In addition, VOAc has been successfully block-copolymerized by combining ATRP with other polymerization processes.151,152 ATRP of halogenated alkenes have not yet been reported in detail. Ring-opening polymerization has been successful for several monomers, especially for those with radical-stabilizing substituents. Potential copolym- erization of these monomers will lead to vinyl poly- mers with a hydrolyzable linkage in the main chain.26,153 Some examples of other monomers (co)- polymerized by ATRP monomers are shown in Scheme 7. In summary, a variety of monomers have been successfully polymerized under ATRP conditions to yield well-defined polymers. For a monomer to un- dergo ATRP, it is important to have stabilizing groups (e.g., phenyl or carbonyl) adjacent to the carbon radicals that produce a sufficiently large atom transfer equilibrium constant but do not interfere with the growing radical and the catalytic system. In addition, it is necessary to adjust the reaction conditions (concentrations, temperature, catalyst) to obtain a suitable radical concentration for a specific monomer. 8. Summary and Outlook ATRP has been successful in controlling polymer- ization of many styrenes, acrylates, and methacry- lates and several other relatively reactive monomers such as acrylamides, vinylpyridine, and acrylonitrile. Atom Transfer Radical Polymerization Chemical Reviews, 2001, Vol. 101, No. 9 2931

However, there are two major classes of monomers which have not yet been successfully polymerized by ATRP. Acidic monomers fail since they can protonate ligands and form the corresponding carboxylate salts. There has been progress in this area, and methacrylic acid in the neutral form of the sodium salt has been polymerized. A similar approach has been reported for other acidic monomers. In principle, use of less basic ligands (oxygen and sulfur-based), which would also complex strongly, may prevent loss of ligands. Additionally, the acids may be used as ligands themselves: iron succinates or halides were reported as ATRP catalysts. Halogenated alkenes, alkyl-substituted olefins, and vinyl esters are presently resistant to polymerization by ATRP. They belong to a class of monomers with very low intrinsic reactivity in radical polymerization and radical addition reactions and could have a very low ATRP equilibrium constant. To polymerize them, it will be necessary to use catalysts with very high reactivity and generally a very negative reduction potential, but this may be accompanied by reduction of the free radicals to carbanions and formation of organometallic species. Such species may react by a coordination pathway rather than via free radical intermediates. Those species may be also hydrolyti- cally less stable and catalysts may be very sensitive to oxygen. The range of monomers polymerizable by ATRP is greater than that accessible by nitroxide-mediated polymerization, since it includes the entire family of methacrylates. However, degenerative transfer pro- cesses, with the RAFT method being currently most often used, allows polymerization of more monomers than ATRP. Perhaps new ATRP catalysts may al- leviate this problem. However, it must be stressed that each group of monomers may be best suited to a specific mechanism. For example, isobutene and vinyl ethers best fit cationic polymerization, R-olefins and perhaps dienes coordination, and/or anionic polymerization, whereas polar monomers such (meth)- acrylates seem to fit the free radical mechanism best. D. ATRP Initiators As discussed previously, the amount of the initiator in the ATRP determines the final molecular weight of the polymer at full monomer conversion. Multi- functional initiators may provide chain growth in several directions (cf. section III.C). Fast initiation is important to obtain well-defined polymers with low polydispersities. A variety of initiators, typically alkyl halides, have been used successfully in ATRP. Many different types of halogenated compounds are poten- tial initiators and are discussed below based on their structure. 1. Halogenated Alkanes Halogenated alkanes, such as CHCl3 or CCl4, are typically used in atom transfer radical addition and were among the first studied as ATRP initiators.15,16 In the ruthenium-catalyzed ATRP of MMA, molecu- lar weights of the polymer increased linearly with the conversion however, at high monomer conver- sion, the molecular weight deviated from the theo- retical values.124 The polymers obtained were mono- modal with low polydispersities (ca. 1.3). In contrast, di- or monochloromethanes were not able to polymer- ize MMA under similar conditions.120 CCl4 has also been used in other catalytic systems, including the Cu-based one.18 When CuCl(bpy)3 was used as the catalyst for the ATRP of styrene at 130 ��C, CCl4 was found to act as a bifunctional initiator.121 Again, deviation of the molecular weights from the theoretic values was observed, and this was tenta- tively explained by additionally generated chains resulting from the activation of the central dichlo- romethylated moiety which undergoes -scission.121 Control of the molecular weight is possible using CHCl3 for the CuCl(bpy)3 system, whereas di- and monochloromethanes lead to uncontrolled polymer- izations.18 In homogeneous systems, CCl4 is some- times less efficient due to a potential outer-sphere electron-transfer (OSET) reaction and the reduction of the radicals to anions (cf. section II.I). Slow addition of the catalyst to the initiating system apparently improves the initiation efficiency.42 With CCl4 and Ni{o,o���-(CH2NMe2)2C6H3}Br as the catalyst, the experimental molecular weight of PMMA in- creased with monomer conversion but showed devia- tion at high conversions,113 similar to the ruthenium system.15 Deviation of molecular weight was also observed for the FeCl2(PPh3)2 catalytic system.116 CCl3Br successfully initiated the controlled polym- erization of MMA catalyzed by RuCl2(PPh3)3,83 NiBr2- (PPh3)2,44 NiBr2(PnBu3)2,114 or Ni(PPh3)4.154 However, with the Ni(II)/(PPh3)2 system, other combinations of initiators and catalysts, such as CCl3Br/NiCl2- (PPh3)2, CCl4/NiBr2(PPh3)2, or CCl4/NiCl2(PPh3)2, re- sulted in bimodal molecular weight distributions at high MMA conversions.44 2. Benzylic Halides Benzyl-substituted halides are useful initiators for the polymerization of styrene and its derivatives due to their structural resemblance. However, they fail Scheme 7. Miscellaneous Monomers (co)Polymerized by ATRP 2932 Chemical Reviews, 2001, Vol. 101, No. 9 Matyjaszewski and Xia

in the polymerization of more reactive monomers in ATRP such as MMA. For example, using CuCl- (dNbpy)2 as the catalyst, inefficient initiation was observed when 1-phenylethyl chloride was employed as the initiator for the polymerization of MMA.123 PMMA with much higher molecular weights than the theoretic values and high polydispersities (Mw/Mn ) 1.5-1.8) were obtained. In contrast, a well-controlled polymerization was realized with benzhydryl chloride (Ph2CHCl) as the initiator under similar conditions. In fact, the radical generation was so fast that slow addition of benzhydryl chloride was necessary to avoid a significant contribution of irreversible biradi- cal termination early in the polymerization.123 Im- provement of the initiation efficiency for the ATRP of MMA using primary and secondary benzylic ha- lides is possible by employing the halogen exchange concept.155 Polyhalogenated benzylic halides have been used for the ATRP of MMA catalyzed by RuCl2(PPh3)3/Al- (OiPr)3.120 PMMA with very low polydispersities were obtained when Ph2CCl2 was used as the initiator. In contrast, PhCCl3 led to a bimodal molecular weight distribution consisting of two narrowly distributed fractions, the higher of which was double the molec- ular weight of the other.120 PhCHCl2 has been also used in Cu-based ATRP of styrene and MMA, appar- ently providing two-directional growth of the poly- meric chains.156 Scheme 8 illustrates some examples of halogenated alkanes and benzylic halides used successfully in ATRP. 3. R-Haloesters Various R-haloesters have been successfully em- ployed to initiate well-controlled ATRP. In general, R-haloisobutyrates produce initiating radicals faster than the corresponding R-halopropionates due to better stabilization of the generated radicals after the halogen abstraction step. Thus, slow initiation will generally occur if R-halopropionates are used to initiate the polymerization of methacrylates. In con- trast, R-bromopropionates are good initiators for the ATRP of acrylates due to their structural resem- blance. In their search for better initiators in ruthenium- mediated ATRP, Sawamoto et al. examined three R-bromoesters of different structures (Scheme 9).120 The malonate with two geminal esters generates radicals faster than 2-bromoisobutyrate and leads to lower polydispersities. The dimeric model of the dormant chain end (dimethyl 2-bromo-2,4,4-trimeth- ylglutarate) initiates a faster polymerization and provides PMMA with lower polydispersities than R-bromoisobutyrate, likely due to the back strain effect 54,157,158 the release of the steric strain of the dormant species during rehybridization from the sp3 to the sp2 configuration leads to a higher equilibrium constant. The dimeric model has also been used in the ATRP of MMA catalyzed by NiBr2(PPh3)2,115 and the chloride analogue of the dimeric model compound leads to the controlled polymerization of MMA and styrene mediated by a half-metallocene-type ruthe- nium complexes.159 Malonate derivatives are less efficient in Cu-based ATRP, perhaps due to the previously mentioned OSET process. Slow addition of the catalyst to the initiator solution in monomer improves control tre- mendously.42 R-Haloesters with various functional groups at- tached can easily be prepared through a straightfor- ward esterification reaction of the appropriate acid halides. Since ATRP can tolerate various functional groups, well-defined end-functional polymers have been conveniently prepared without the need for additional protecting reactions. A variety of function- alities, such as hydroxy, epoxy, allyl, vinyl, ��-lactone, and carboxylic acid have been introduced onto the R-end of the polymer by use of a functional initiator and will be discussed in later sections (Scheme 10).99,128,160 Polyhalogenated R-haloesters (e.g., CCl3CO2CH3 and CHCl2CO2CH3) have also been successfully ap- plied as initiators for the ATRP of MMA catalyzed by RuCl2(PPh3)3/Al(OiPr)3.120 Multiarm stars of PMMA are produced when multifunctional dichloroacetates are used in the ruthenium-catalyzed ATRP.161,162 Mixed benzyl and ester derivatives such as methyl R-bromophenylacetate were successfully used in the aqueous polymerization of 2-(dimethylamino)ethyl methacrylate.163 4. R-Haloketones An R-bromoketone has been used to initiate the controlled polymerization of MMA catalyzed by Ni{o,o���-(CH2NMe2)2C6H3}Br113 and Ni(PPh3)4.154 Polyhalogenated R-haloketones (e.g., CCl3COCH3 and CHCl2COPh) are among the best initiators Scheme 8. Some Halogenated Alkanes and Benzylic Halides Used as ATRP Initiators Scheme 9. Various r-Bromoesters Used in Ruthenium-Mediated ATRP of MMA Scheme 10. Representative Functional Initiators Derived from r-Haloesters Atom Transfer Radical Polymerization Chemical Reviews, 2001, Vol. 101, No. 9 2933

for the ATRP of MMA catalyzed by ruthenium complexes.83,120,159,164,165 Well-controlled polymers with low polydispersities (Mw/Mn 1.20) have been ob- tained. The stronger electron-withdrawing power of the ketone���s carbonyl induces further polarization of the carbon-chlorine bond, which is attributed to the faster initiation observed with the ketones than with the ester counterparts. 5. R-Halonitriles R-Halonitriles are fast radical generators in ATRP, due to the presence of the strong electron-withdraw- ing cyano group. Moreover, the radical formed after halogen abstraction is sufficiently reactive, which leads to fast initiation through rapid radical addition to monomer. Of the initiators studied for the polym- erization of acrylonitrile catalyzed by copper com- plexes, 2-bromopropionitrile resulted in polymers with the lowest polydispersities.131 2-Bromopropioni- trile is also the initiator of choice when a bromine initiator is desired in the iron-mediated ATRP of MMA.98 However, R-halonitriles were not used in ruthenium-catalyzed ATRP as the cyano group de- activates the catalyst by forming a strong complex with ruthenium.120 6. Sulfonyl Halides As ATRP initiators, sulfonyl chlorides yield a much faster rate of initiation than monomer propagation.55 The apparent rate constants of initiation are about four (for styrene and methacrylates) and three (for acrylates) orders of magnitude higher than those for propagation. As a result, well-controlled polymeriza- tions of a large number of monomers have been obtained in copper-catalyzed ATRP.19,55 End-func- tional polymers have been prepared using sulfonyl chlorides where functionalities were introduced onto the aromatic ring.166 The phenyl group substituent has only a small effect on the rate constant of initiation because the sulfonyl radical and its phenyl group are not related through conjugation. A unique feature of the sulfonyl halides as initia- tors is that although they are easily generated, they only dimerize slowly to form disulfones and slowly disproportionate. Thus, they can react with the monomers and initiate the polymerization effi- ciently.167 When sulfonyl chlorides were used in the polym- erization of MMA catalyzed by RuCl2(PPh3)3/Al- (OiPr)3, S-shaped conversion vs time profiles were obtained.168 Moreover, experimental molecular weights were higher than the theoretical values, indicating a low initiator efficiency. The polydispersities were around 1.2-1.5. The low initiator efficiency was explained by the formation of sulfonyl esters from sulfonyl chlorides and Al(OiPr)3 during the early stages of the polymerization. Examples of sulfonyl chlorides used as ATRP initiators are shown in Scheme 11. 7. General Comments on the Initiator Structure in ATRP Two parameters are important for a successful ATRP initiating system. First, initiation should be fast in comparison with propagation. Second, the probability of side reactions should be minimized. Analogous to the ���living��� carbocationic systems, the main factors that determine the overall rate con- stants are the equilibrium constants rather than the absolute rate constants of addition.169,170 There are several general considerations for the initiator choice. (1) The stabilizing group order in the initiator is roughly CN C(O)R C(O)OR Ph Cl Me. Multiple functional groups may increase the activity of the alkyl halide, e.g., carbon tetrachlo- ride, benzhydryl derivatives, and malonates. Tertiary alkyl halides are better initiators than secondary ones, which are better than primary alkyl halides. These have been partially confirmed by recent mea- surements of activation rate constants.171-173 Sulfonyl chlorides also provide faster initiation than propaga- tion. (2) The general order of bond strength in the alkyl halides is R-Cl R-Br R-I. Thus, alkyl chlorides should be the least efficient initiators and alkyl iodides the most efficient. However, the use of alkyl iodides requires special precautions. They are light sensitive, can form metal iodide complexes with an unusual reactivity (e.g., CuI2 is thermodynami- cally unstable and cannot be isolated), the R-I bond may possibly be cleaved heterolytically, and there are potential complications of the ATRP process by degenerative transfer.174,175 By far, bromine and chlorine are the most frequently used halogens. In general, the same halogen is used in the initiator and the metal salt (e.g., RBr/CuBr) however, the halogen Scheme 11. Examples of Sulfonyl Chlorides used as ATRP Initiators 2934 Chemical Reviews, 2001, Vol. 101, No. 9 Matyjaszewski and Xia

exchange can sometimes be used to obtain better polymerization control.155 In a mixed halide initiating system, R-X/Mt-Y (X, Y ) Br or Cl), the bulk of the polymer chains are terminated by chlorine due to the stronger alkyl-chloride bond. Thus, the rate of initiation is increased relative to propagation and ethyl 2-bromoisobutyrate/CuCl leads to a better- controlled polymerization of MMA in comparison to using ethyl 2-bromoisobutyrate/CuBr.155 A similar result has also been observed in Ru-based ATRP.176 The halogen exchange method also enables the use of alkyl halides of apparently lower reactivities in the polymerization of monomers with apparently higher equilibrium constants. This is especially important for the formation of block copolymers.177-180 Pseudo- halogens (e.g., SCN) have also been used in ATRP.29,33 Initiation using benzyl thiocyanate is slow for both styrene and MA, and Mn higher than the theoretical values are obtained. Better results are obtained when alkyl halides are used as the initiators and CuSCN as the catalyst. Similarly, transition metal dithio- carbamates have been employed in the presence of AIBN to induce controlled reverse ATRP of styrene at 120 ��C. Good agreement between theoretical and experimental Mn values were obtained with Mw/Mn ) 1.15-1.30.32 (3) Successful initiation in ATRP can depend strongly on the choice of catalyst. For ex- ample, 2-bromoisobutyrophenone initiates the con- trolled polymerization of MMA catalyzed by ruthe- nium or nickel complexes but has not been successfully used in the copper-mediated ATRP. This is ascribed to the reduction of the resulting electrophilic radical by the copper(I) species as the copper catalysts have lower redox potentials. (4) The method or order of reagent addition can be crucial. For example, slow addition of the benzhydryl chloride initiator to the CuCl(dNbpy)2-catalyzed ATRP of MMA generates a lower concentration of benzhydryl radicals and thus reduces the rate of termination between the radicals. The diethyl 2-bromomalonate/CuBr system initiates the ATRP of styrene, and the polymerization was well controlled when the catalyst was added slowly to the initiator/monomer solution. This avoided the poten- tial reduction of the malonyl radical by the copper(I) species. It may also be surprising, but the heteroge- neous catalytic systems may provide more efficient initiation than homogeneous ones when very reactive alkyl halide initiators are used, most likely due to slow dissolution of the catalyst and hence its lower instantaneous concentration. For example, CCl4 is a good initiator for styrene and MMA with CuBr(bpy)3 as the catalyst,18 but the same is not true using the CuBr(dNbpy)2 catalytic system. The initiation ef- ficiency increased when the catalyst solution was added slowly to the initiator solution.42 8. Summary and Outlook Range of available initiators for ATRP is much larger than for other CRP methods. In fact, many NMP and RAFT reagents are prepared from ATRP initiators, i.e., activated alkyl halides by either nu- cleophilic displacement (RAFT) or radical trapping in the presence of Cu(0) (NMP). The basic require- ment for a good ATRP initiator is that it should have reactivity at least comparable to that of the subse- quently formed growing chains. This also indicates that not all initiators are good for all monomers. This is an extremely important criterion for the prepara- tion of block copolymers. Very reactive initiators may produce too many radicals, which will terminate at early stages. This will reduce efficiency of initiation, produce too much of the deactivator, and slow the process. Perhaps one of the few exceptions is a class of sulfonyl halides which terminate relatively slowly by the irreversible radical termination. It is necessary to better correlate structures of the alkyl halides with their ATRP reactivities. This includes both the alkyl part (electronic and steric effects) and (pseudo)halogens. The ATRP reactivity includes not only the BDE of the C-X bond, but also halogenophilicity of the transition metal. Thus, the structure-reactivity correlation should include both components as well as the effects of solvent and temperature. Comparison of model and macromo- lecular compounds is also important as well as extension to dense systems to compare intra- and intermolecular effects. This will be especially impor- tant for the macromolecular engineering of complex polymeric structures. Halogen end groups are an inherent part of the ATRP systems. They can be replaced by many synthetic methods to provide more useful function- alities and provide halogen-free products. Pseudohalo- gens such as (iso)thiocyanate and azide groups have also been used as exchangeable end groups in ATRP and are quite attractive, since they may be hydro- lytically more stable and can provide direct pathways to end-functional polymers. There are many multi- functional activated halides which enable simulta- neous growth of chains in several direction, leading to star, comb, and brush macromolecules. E. Transition Metal Complexes A number of transition metal complexes have been applied in ATRP. As mentioned previously, to gener- ate growing radicals, the metal center should undergo an electron transfer reaction with the abstraction of a (pseudo)halogen and expansion of the coordination sphere. In addition, to differentiate ATRP from the conventional redox-initiated polymerization and in- duce a controlled process, the oxidized transition metal should rapidly deactivate the propagating polymer chains to form the dormant species. The applications and scope of the different transition- metal complexes are discussed following their peri- odic groups. 1. Group 6: Molybdenum and Chromium A series of lithium molybdate(V) complexes [LiMo- (NAr)2(C-N)R] (C-N ) C6H4(CH2NMe2)-2 R ) (C- N), Me, CH2SiMe3, or p-tolyl), have been used in the ATRP of styrene using benzyl chloride as the initiator (Scheme 12).181 The molybdate(V) complexes were generated in situ from the reaction of the correspond- ing molybdenum(VI) complexes [Mo(NAr)2(C-N)R]. Relatively high polydispersities (Mw/Mn ��� 1.5) were obtained, and the efficiency of the benzyl chloride initiator was rather poor (6-18%), which was as- cribed to the extreme air-sensitivity of the lithium Atom Transfer Radical Polymerization Chemical Reviews, 2001, Vol. 101, No. 9 2935