New Crosslinking Processes BY GBRARD BEINERT, AHMED BELKEBIR-MRANI, JEAN HERZ, * GBRARD HILD AND PAUL REMPP Centre de Recherches sur les MacromolCcules-CNRS- 6, rue Boussingault-67083 Strasbourg Cedex (France) Received 2nd January, 1974 Synthesis of model-networks characterized by the quasi-constant length of the linear chain elements between two successive branch points has been carried out by anionic block copolymerization of two monomers, one being bifunctional. The influence of various parameters (concentration, temperature, number of molecules of bifunctional monomer added per active chain end) on the behaviour of the gels is discussed. This method can be applied to several systems. Another method was successfully used to synthesize networks : reaction of a, w-difunctional linear polymer chains with tri- or tetrafunctional molecules, in stoichiometric amount was shown to lead to gels in which both the length of the linear chain elements and the functionality of the branch points are controlled. Reactions of terminal carbon metal bonds with electrophilic groupings of various kinds were used, as well as reaction of silane end groups with allylic double bonds.The gels obtained were characterized by their swelling behaviour, in relation to the molecular weight of the chain elements and the functionality of the branch points. The numerous methods which have been developed to synthesize crosslinked polymeric networks can be classified into two groups :-(i) Methods involving random copolymerization of two monomers, one of which is bifunctional. (ii) Methods involving bridge formation between preexisting linear polymer chains, by vulcaniza- tion, peroxidation, etc.Neither of these methods yields well defined networks, since copolymerization as well as chemical transformation of polymers are random processes. The networks obtained cannot be characterized easily by their structural parameters. The length of their linear chain elements fluctuates very much around its average value, which is not experimentally accessible. In many cases the homogeneity of the gels is not satisfactory, especially when syneresis (solvent expulsion) takes place during the process. Finally the “ gel point ” has no physical significance since it usually takes place at an early stage of the reaction. The present paper gives an account of research on the synthesis of well-defined model-networks, by several methods.A model-network is a cross-linked macro- molecule which consists of v elastic chain elements connected by (2v/’) $functional branch points. The linear chain elements are in first approximation identical in length and each of them is connected by its two ends to two diflerent branch points. The model-network should be homogeneous, it should contain neither pendent chains nor loops, and no solvent expulsion should take place during its preparation. Finally the gelation process itself should not involve any major change of the segment concentration; this is a necessary condition for the gel point to have a physical significance. One can expect, therefore, to establish a relationship between the segment concentration at the gel point and the “ memory-term ” h, characterizing the relaxation state of the linear chain elements at the segment concentration at which crosslinking occurred.2728 NEW CROSSLINKfNG PROCESSES The principle of these new crosslinking processes is quite different from the above indicated classical methods. Here a linear " precursor " polymer is equipped with reactive end-groups ; in a second step branch points are formed, each connecting f chain-ends together. This method allows one to characterize adequately the linear " precursors " molecules, and to check for their polydispersity. Furthermore, their average molecular weight can often be chosen rather precisely in advance, so one can choose the porosity of the network, since the length of the elastic chains determines the average pore size of the network. Three methods will be discussed in this paper.Two of them are anionic cross- linking methods, the third one starts from a, o-difunctional polymeric chains which are reacted under proper conditions with a plurifunctional reagent. NETWORK FORMATION BY ANIONIC BLOCK COPOLYMERIZATION POLYSTYRENE MODEL-NETWORKS It was shown '* a few years ago that star-shaped polystyrene can be prepared anionically by block copolymerization of styrene and some divinylbenzene, the initiator being monofunctional : butyl-lithium or cumyl-potassium (fig. 1). The reaction is carried out at low temperature, in an aprotic solvent, under inert atmos- phere. STAR POLYMER FORMATION FIG.1. In a first step, a monofunctional " living " polystyrene is obtained by reaction of the initiator with a given amount of styrene. Such a polymer exhibits a sharp molecular weight distribution because anionic " living " polymerization with fast initiation yields macromolecules of low polydispersity. To the solution of this " living " polymer a small amount of divinyl-benzene (DVB) is added : its poly- merization is initiated solely by the living carbanionic sites. Each star-molecule is constituted by a smalf crosslinked nodule of poly@VB) connected with f identical linear polystyrene branches. These " star '' molecules can be well characterizedG . BEINERT, A. BELKEBIR-MRANI, J . HERZ, G. HILD, P. REMPP 29 by the length of the individual branch (which is the “ precursor ” polystyrene itself) and byfwhich can be determined by the ratio of the molecular weight of the “ star ” molecule to that of the “ precursor ”.such as bi-sodium a-methylstyrene ‘‘ tetramer ” or naphthalene-sodium, a bifunctional “ living ” polystyrene is formed, the chain of which is equipped at both ends with organometallic sites. Addition of a small amount of DVB to the solution of this “ living ” polymer leads to rapid gelation of the reaction medium. A crosslinked network is formed in which each linear chain element (“ precursor ”) should be con- nected with two different branch points (poly-DVB nodules). Here again, the average length of the “ precursor ” polystyrene can be chosen arbitrarily, and determined precisely. If the experimental conditions have been chosen adequately (solvent, concentration, temperature, adequate stirring etc.) the obtained gels are homogeneous and close to ideality.It should be noted that in such a process, gelation occurs without any major change of the segment concentration, since it only involves polymerization of the small amount of DVB initiated by the “living” ends of the “ precursor ”. The only drawback of this method of synthesis is that the actual average functionality fof the branch points (number of elastic chains connected with a given nodule of poly-DVB) remains unknown : f is neither determined, nor experimentally accessible. The polystyrene model-networks so obtained have nevertheless been successfully 5* used to test the validity of equilibrium swelling theories on ideal gaussian networks.We have been able to conclude that this model of the ideal gaussian network, as developed by FloryY2’ and more recently by Dusek and Prins fits satisfactorily. This result indicates that defects such as pendent chains, couplings, loops, and also entanglements are few, and that the behaviour of the elastic chains in the swollen network follows gaussian statistics. Moreover, from equilibrium swelling measure- ments in a pure diluent on a homologous series of polystyrene networks-using precursor polymers of various molecular weights-it was shown that the experimental results were consistent with a functionality of the branch points of the order of 4. If instead of a monofunctional initiator, we use a bifunctional initiator GENERALIZATION The preceding method of network formation by anionic block copolymerization has been applied to other anionically polymerizable monomer systems.In such a process, it is necessary that the carbanionic “living” ends of the “precursor” polymer initiate rapidly and quantitatively the polymerization of the bifunctional monomer. This condition involves that the electroaffinity of the bifunctional monomer be equal to or greater than that of the first monomer. In the following experiments, the crosslinking agent is either divinylbenzene (DVB) or ethylene dimethacrylate (DME) : CH2=C-C-O-CH,-CH2-O-C-C=CH, II I 0 CH3 I II CH30 Model-networks have been obtained with the following systems : styrene + DVB ; isoprene + DVB ; 2-vinylpyridine + DVB (or DME) * ; methylmethacrylate + DME * ; butylmethacrylate + DME ; isopropylidene glyceryl methacrylate -I- DME.S30 NEW CROSSLINKING PROCESSES These results call for the following comments : (a) Concerning gels of plyisoprene it would have been interesting to obtain 1,4 cis stereoregular plyisoprene chain elements between branch points in order to compare the properties of the gels obtained with vulcanized rubber.But it is well known that 1,4 cis polymerization of isoprene occurs only in non-polar solvent media. Unfortunately, none of the common bifunctional Li-initiators is soluble in non-polar solvents. Recent attempts to over- come this difficulty are rather promising, however. (b) With poly-2-vinylpyridine gels, curiously enough, DVB can be used as the second monomer, which means that the electroaffinity of DVB is definitely higher than that of styrene.(c) For metha- crylic esters, DVB cannot be used since its polymerization cannot be initiated by the polymethacrylate ester anions. In this case, DME is used as second monomer. (d) Water soluble gels cannot be prepared anionically. But it has been possible to synthesize isopropylidene glyceryl methacrylate monomer : / \ CH3 CH3 and to obtain anionic gels using ethylene dimethacrylate as bifunctional monomer. In this case, subsequent acid hydrolysis of the networks obtained destroys the acetal group and regenerates the remaining OH functions of glycerol, without touching the ester function ; this process yields polyglyceryt methacrylate networks, CH2=C-C-O-CH2-CH-CH2 I I OH OH I II CH30 glyceryl met hacrylate which swell in water.NETWORK FORMATION BY CHEMICAL REACTIONS INVOLVING " LIVING " ANIONIC POLYMERS It is well known that anionic " living '' polymers react not only with any proton donating substance, but also with many electrophilic functions, to yield a-functional polymers. Alkyl halides, acid chlorides, esters, nitriles, anhydrides, as well as carbon dioxide or ethylene oxide do react with carbanionic sites. The idea of using pluri- functional electrophilic reagents to link two or more chain-ends together has been used many times. Coupling agents as COC1, or (CH3),SiC12 are commonly used. Star polymer synthesis was attempted ClCH, CH,CI using various compounds such as : but two types of difficulties were encountered: (i) the decreasing reactivity of the sites, as the degree of substitution increases and (ii) metal halogen exchange reactions, which yield undesired side products.In most cases using monocarbatiionic precursor polymers, mixtures of star- molecules with p , p - 1, p - 2 . . . branches were obtained. Moreover, an excess ofG . BEINERT, A . BELKEBIR-MRANI, J . HERZ, G . HILD, P. RBMPP 31 fiving polymer has generally to be used to attain complete reaction. Therefore, one cannot expect to obtain adequate model-networks by using this method with bifunctional precursors. Many structural defects would be introduced in the molecular structure during the process. But it was noticed that tris(ally1oxy)triazine (TT) reacts quantitatively with “ living ” monocarbanionic polystyrene to yield star-molecules according to the reaction ’ : s with R = CH2=CH-CH2-- This shows that the three functions do react.The same reaction can be applied to synthesis of model-networks. Thus, starting from bifunctional “ living ” polystyrene and reacting it with a stoichiometric amount of TT, under efficient stirring, one obtains well-defined networks. This method enables one to synthesize networks in which both the length of the elastic chains and the functionality of the branch points are known, and are subject only to very small fluctuations within a sample. It has to be assumed, however, that no pendent chains are left over (which is a reasonable assumption if stoichiometry and stirring are adequate) and that only very few loops are formed. As a matter of fact, the probability of formation of such loops can be calculated roughly, it depends both upon the molecular weight of the precursor and upon the overall concentration at the gel point.It should also be emphasized that the gelation does not involve any variation of the segment concentration, since the crosslinking process merely consists of bond formation between chain-ends of several molecules. Attempts were made to extend this method to preparation of networks with tetra (or even hexa) functional branch points. To achieve this, a new crosslinking agent was made by reacting an excess of TT with tetraphenyl-disodiobutane. The adduct obtained, bis(diallyloxytriazy1) tetraphenylbutane, is purified and can be used as tetrafunctional deactivator for living polystyrene.All0 OAll This was first checked again with monocarbanionic polystyrene. As expected the obtained star-polymer has four branches. Next, a bifunctional living polystyrene was reacted in stoichiometric amount with this compound. Gelation occurs as expected, while the red colour of the carbanions vanishes. The model-network thus obtained is constituted of elastic chains connected by tetrafunctional branch points ;32 NEW CROSSLINKING PROCBSSBS if stoichiometry has been effective and stirring adequate, one can expect that the number of structural irregularities will remain low. NETWORK FORMATION BY CHEMICAL REACTIONS INVOLVING FUNCTIONS AT CHAIN ENDS As already mentioned, chemical reactions on a polymer chain is one effective method for the production of tridimensional networks.The most famous of these reactions is the so-called vulcanization of rubber. But, aside from the fact that such a reaction is a random process, it should be indicated that the number of pendent chains in such a process amounts to almost twice the number of primary molecules, since there is little chance that the crosslinking reaction occurs precisely at chain ends. Such networks are, therefore, far from ideal ; they cannot be considered to be well-defined. Recently, much attention has been devoted to network synthesis by reactions involving functions located at chain ends. The carbanionic sites of " living " polymers can play the role of active functions, as was shown in the preceding section. But several other systems have also been investigated, especially crosslinking of a, o-dihydroxy polymers (or oligomers) using commercial plurifunctional urethanes as crosslinking agents.We have used a similar method of making polymeric networks, but we have used a quite different reaction, namely the addition of silanes onto allylic double-bonds, according to l9 1 I I I - Si-H + CH,=CH-CH,-R-+ - Si-CH2-CH2-CH2-R CH3 CH3 We attempted to react polydimethylsiloxanes fitted at chain ends with silane functions * onto tetrallyloxyethane.20 The reaction is carried out either in bulk or in the presence of a diluent : heptane, a good solvent of the PDMS chain, or toluene, a poor solvent of the polymer. Stoichiometric amounts of the reactants are needed as well as adequate mixing to provide satisfactory homogeneity of the reaction mixture.Chloroplatinic acid catalyses the reaction, which takes place at tempera- tures ranging from 20" to 70°C. Gelation of the reaction mixture takes place readily, and the model-networks thus obtained are constructed from elastic PDMS chains connected by their ends with tetrafunctional branch-points. To obtain adequate stoichiometry, precise characterization of the PDMS- precursor molecules is necessary. Molecular weight measurements are performed by osmometry, as well as by silane end group determination. The agreement is satis- factory. The polydispersity of the samples should not be too broad, though molecular weight average measurements by light scattering are possible only for the highest molecular weight samples. Also some gel-permeation chromatography diagrams were obtained.So far as the imperfections of the networks are concerned, the same remarks can be made as in the preceding case: pendent chains may be left over, Ioops may be formed, and on the other hand chain entanglements may lead to some supplementary " physical " crosslinks. * The PDMS fitted with silane end groups were obtained from the Silicon Division of Rhone- Poulenc ; we wish to express our appreciation for its help in this work.@. BEINERT, A . BELKEBIR-MRANI, J . HERZ, G . H I L D , P. REMPP 33 The crosslinked samples were submitted to prolonged solvent extraction (Soxhlet). It was found that the gels contained only very little extractable polymer. It should be noted, in addition, that the PDMS precursor samples may well contain a small percentage of cyclic macromolecules which obviously cannot take part in the reaction.This shows that the probability for any reactive chain end to react is very close to 100 %, and from this result it follows that the percent amount of pendent chains should remain very low. We attempted to use a similar method to make PDMS networks with branch points of functionality other than 4. Triallyloxytriazine was tried as a 3-functional reagent. In that case a network is formed at first, as expected, but it is again destroyed on standing for a couple of days, because the acid which catalyses the reaction is able to hydrolyse the triazine links between chain ends : the precursor is thus reformed. A six-functional reagent was also synthesized : bis (triallyoxy methyl) ethyl ether : This compound was obtained from dipentaerythritol and ally1 bromide as starting materials.This hexa-functional reagent yields nice crosslinked model-networks with our PDMS precursor molecules. In order to relate the physical behaviour of the PDMS gels to their structural characteristics we have carried out equilibrium swelling and uniaxial compression experiments. Our results indicate that within the limit of experimental accuracy the crosslinked gels obtained show ideal behaviour according to theoretical expecta- t i ~ n . ~ Of course, this does not mean that no defects (pendent chains, loops, entanglements) are present in the gel, but their total effect on its physical behaviour is negligible. CONCLUSION Until recently, any theoretical or experimental approach to the physical and mechanical behaviour of crosslinked gels and vulcanizates required assumptions concerning their structural characteristics ; the necessity of creating well-defined model-network syntheses was evident.The purpose of our paper is to show that several methods are now available to prepare such networks, and to report results obtained recently in this field in our laboratory. Comparison between theoretical expectation and experimental results obtained with model gels yielded rather satisfactory agreement, showing the validity of the main hypothesis made in the theoretical treatment of Dusek and Prins. However, it should be emphasized once more that the limitation of the preparation methods cannot be overcome entirely, since structural defects such as pendent chains, loops, entanglements and fluctuations over M and fare still possible to some extent.Our experimental results however have shown that their influence can be neglected in first approximation. J. G. Zilliox, P. Rempp and J. Parrod, J. Polymer Sci. C, 1968,22, 145. D. J. Worsfold, J. G. Zilliox and P. Rempp, Canad. J. Chem., 1969,47, 3379. A. Kohler, J. Polacek, I. Kossler, J. G. Zilliox and P. Rempp, European Polymer J., 1972,8,627, P. Weiss, G. Hild, J. Herz and P. Rempp, Makromol. Chem., 1970, 135, 249. A. Haeringer, G. Hild, P. Rempp and H. Benoit, MakromoZ. Chem., 1973,169,249. D. Froelich, D. Crawford, T. Rozek and W. Prins, Macromolecules, 1972, 5, 100. ’ K. Dusek and W. Prins, Adu. PoZyrner Sci., 1969, 6, 1. * G. Hild and P. Rempp, Compt. Rend., 1969,269, 1622. G. Beinert, G. Hild and P. Rempp, Makromol. Chem.. 1974, 175, 2069. l o S. P. S. Yen, Makromol. Chem., 1962, 81, 152. 57-B34 NEW CROSSLINKING PROCESSES l 1 T. Altares, D. P. Wyman, V. R. Allen and K. Meyersen, J. Polymer Sci. A , 1965, 3, 4131. l2 W. A. Brice, G. M. Gibbon and I. E. Meldrum, Polymer, 1971,11,290. l3 J. C. Meunier and R. van Leemput, Makromol. Chem., 1971,142,l. l4 J. E. L. Roovers and S . Bywater, Macromolecules, 1972,5, 384. l5 M. Morton, T. H. Helminiak, S. Gadkary and F. Bueche, J. Polymer Sci., 1962,57,471. l6 J. A. Gervasi and A. B. Gosnell, J. Polymer Sci. A-1, 1966,4, 1391. l 8 M. Hert, C1. Strazielle and J. Herz, Compt. Rend. C, 1973, 276, 395. l9 G. Greber, Angew. Makromol. Chem., 1971,141,145. 2o J. Herz, A. Belkebir-Mrani and P. Rempp, Eurupean Polymer J., 1973, 9, 1165. 21 P. J. Flory, Principles of polymer chemistry (Cornell University Press, Ithaca, N.Y., 1953). J. Herz, M. Hert and C1. Strazielle, Makromol. Chern., 1972, 160,213.