7 Are Cured Thermosets Inhomogeneous? 7.1 INTRODUCTION 7.1.1 Why This Question? Compared with thermoplastics (TP), the morphology of thermosets (TS) has not been studied thoroughly. Most TS are amorphous and composed of highly crosslinked molecular networks. But the concept of homogeneous infinite networks represented by one giant molecule (Chapter 3), has long been questioned. For example, epoxy networks (the most studied networks) are very often reported in the literature as inhomogeneous. Historically, this claim was supported by electron microscopic observations of free and frac- ture surfaces of epoxy networks, revealing the presence of a nodular mor- phology in the range of 10–100 nm. Nodules were suggested to be sites of higher crosslink density, resulting from intramolecular crosslinking and cyclization reactions. This conclusion was supported by the observation that, upon etching, the internodular material was preferentially attacked (Racich and Koutsky, 1976). However, similar structures were observed with etched surfaces of amorphous linear thermoplastics, such as polystyrene and poly(methyl methacrylate). Moreover, the small-angle X-ray scattering (SAXS) spectra of simple epoxy networks based on diepoxy and diamine monomers were not essentially different from those of common amorphous polymers. Therefore, it was concluded that the inhomogeneous structure was not an inherent property of epoxy networks (Dusek et al., 1978). A strong argu- ment that supported the view of homogeneous polymer networks was the excellent agreement between experimental values of statistical parameters (gel conversion; evolution of the mass-average molar mass, M w ,inthe pregel stage and the sol fraction and the crosslink concentration in the postgel stage) with theoretical values arising from mean-field models (Chapter 3). However, this subject continued to be a controversial matter. Some authors used atomic force microscopy (AFM) in tapping mode to ‘‘prove’’ the existence of a two-phase structure in epoxy networks; this structures comprises a hard microgel phase and a dispersed phase of soft partially reacted material (Vanlandingham et al., 1999). However, the interpretation of this kind of experimental result seems still to depend very much on what one wants to find. There have been attempts to relate the assumed nodular morphology with the physical properties of networks (Labana et al., 1971). This point is important, because if crosslinked polymers are considered as homogeneous three-dimensional structures, their ultimate properties can be related to the properties of such a continuum. On the other hand, if they are inhomo- geneous, the supramolecular structure shown in Fig. 7.1 provides a more fruitful approach to interpreting of macroscopic properties. The presence of inhomogeneities in some polymer networks, particu- larly those formed by a free-radical chainwise polymerization, has long been recognized (Labana et al., 1971; Dusek, 1971); however, this does not mean that all thermosets must be inhomogeneous. 7.1.2 What Are Inhomogeneities? It is not an easy task to define inhomogeneities in the structure of a polymer network. Every system will exhibit the presence of defects and fluctuations of composition in space when the scale of observation becomes smaller and smaller. A hierarchy of structures exists, from atomic dimensions to the macroscopic material. A scheme of different scale levels used to describe linear and crosslinked polymer structures is shown in Fig. 7.2. Inhomogeneities described in the literature for polymer networks are ascribed to permanent fluctuations of crosslink density and composition, with sizes varying from 10 nm up to 200 nm. This means that their size lies in the range of the macromolecular scale. Another problem for describing a heterogeneous structure is to define boundaries between inhomogeneities and the rest of the structure. In some Are Cured Thermosets Inhomogeneous? 207 cases boundaries are not defined or are fractals. Small inhomogeneities that are well defined and well dispersed in a matrix can be described as an ordered structure on a larger scale. 7.1.3 How to Characterize Inhomogeneities? The presence of inhomogeneities can be characterized by various physical methods. Study of the scattering of electromagnetic radiation from poly- mers embraces a wide variety of techniques and yields information on diverse intramolecular and intermolecular properties. Methods include X- ray, neutron, and light scattering; some features of the scattering from these diverse sources can be treated with a common formalism. The energy vs wavelength dependence of different types of radiation is shown in Fig. 7.3. X-ray and visible light energy ranges are limited, whereas neutrons obtained from a pulsed source span a vast energy range (Gabrys and Tomlins, 1989). Light, small-angle X-ray or neutron scattering methods can characterize the size, internal structure, and spatial arrangement of inhomogeneities. The detection limit increases as the wavelength, , decreases. The average 208 Chapter 7 FIGURE 7.1 Two-dimensional representation of crosslink density inhomo- geneities. A ‘‘gel ball’’ at the center is loosely tied to six similar surrounding regions. Crosslinking in bulk requires that branch points and chain ends are evenly distributed independently of morphology. (Labana et al., 1971 with permission from Kluwer Academic) diameter of an inhomogeneity has to be in the order of magnitude of =4to be detected. Microscopies offer a more integral response. Other techniques such as thermal and thermomechanical analysis, and methods sensitive to local mobility such as nuclear magnetic resonance (NMR), can also be used. One major problem of all these techniques is the sensitivity in the parameter selected to detect the presence of inhomogeneities. With visible light for example, inhomogeneous samples can appear transparent if the difference in the refractive index between the phases is less than 0.01. Staining (in the case of transmission electron microscopy, TEM), or chemi- cal etching (in the case of scanning electron microscopy, SEM), can be helpful in revealing the structure. Are Cured Thermosets Inhomogeneous? 209 FIGURE 7.2 The different scale levels used to describe linear and crosslinked polymers. Valuable information on whether the structure is homogeneous or inhomogeneous can also be obtained by analyzing the network formation process. A shift of experimental and estimated statistical parameters (M w , gel point conversion, sol fraction, etc.) will be observed if inhomogeneities are formed as a result of the crosslinking process. 7.1.4 Aim of This Chapter The structure of a polymer network at a particular conversion is determined by the structure of the starting components, the initial composition of the system, the reaction paths (mechanism and kinetics), and the network for- mation history (staging). The aim of this chapter is to try to answer the question ‘Are cured thermosets inhomogenous’ by giving some examples of both homogeneous and inhomogeneous systems and by analyzing different types of inhomogeneities and the reasons for their formation. Stepwise (polyaddition and polycondensation) and chainwise polymerizations are considered separately, to analyze the way in which the different mechanisms of network formation can fix eventual inhomogeneities produced along the reaction. The possibility of controlling and taking advantage of the formation of a heterogeneous structure is also analyzed. 210 Chapter 7 FIGURE 7.3 The energy-wavelength dependence for different types of radia- tion: photons, infrared, and neutrons. For neutrons (n), ps denotes the pulsed source and ss the steady source. (Gabrys and Tomlins, 1989 – Copyright 2001 – Reprinted by permission of John Wiley & Sons, Inc.) 7.2 THERMOSETS FROM STEP-POLYMERIZATION MECHANISM 7.2.1 Ideal Networks from Stepwise Polymerizations a. Characterizations As explained in Sec. 7.1, epoxy networks have been and are still the subject of controversy. This is mainly based on the particular interpretation of results obtained using microscopy techniques. On the contrary, results obtained with small-angle neutron scattering (SANS) proved that typical diepoxy–diamine networks were homogeneous (Wu and Bauer, 1985). In addition, from thermal and thermomechanical measurements, it is found that typical epoxy–amine networks exhibit one glass transition tem- perature, T g , and one sharp well-defined relaxation peak. The same tech- niques were used for crosslinked polyurethanes based on triol and diisocyanate or diol and triisocyanate (Andrady and Sefcik, 1983). Similar conclusions to those found for epoxy–amine networks were attained. b. Network Buildup If inhomogeneities are formed as a result of the crosslinking process, the evolution of the i-mer distribution, the gel point conversion, the sol fraction, etc., would be affected (Sec. 3.2.2, A 4 +B 2 stepwise polymerization). For diepoxy–diamine systems, most of the experimental gel point conversions reported in the literature are equal to the theoretical ones within experimen- tal error. They are only determined by the initial composition of the system and the relative reactivities of primary and secondary amino groups (Dusek et al., 1977, Dusek, 1986). One example of the good fit of experimental data to the values calculated assuming homogeneity of the system (Chapter 3) is given in Fig. 7.4. The sol fraction is particularly sensitive to the variations of composition and conversion of reactive groups (Ilavsky et al., 1984). c. First Conclusion Concerning Stepwise Polymerizations It can be stated that networks based on a simple formulation (one monomer reacting with a comonomer), obtained from the step-polymerization process will exhibit a homogeneous structure. This is the case for epoxy–amine networks (the most studied) and polyurethane networks that have been used very often as ideal networks for structure–property correlations. Are Cured Thermosets Inhomogeneous? 211 7.2.2 But the Step-Polymerization Process Can Also Induce Inhomogeneities a. Thermodynamically Induced Inhomogeneities If the monomer and the comonomer are initially miscible, reaction products will usually contribute to increasing the miscibility, because of the similarity of the chemical structures of the different i-mers, as their molar masses increase along conversion. Although this is the general rule, some exceptions have been reported. This is the case of a diepoxy (DGEBA, diglycidyl ether of bisphenol A) reacted with a diamine based on a long polypropylene oxide chain (typically, number-average molar mass, M n ¼ 2000 g mol À1 ). The two monomers are initially miscible but, as conversion progresses, a phase separation of domains (possibly) rich in the polypropylene oxide chains 212 Chapter 7 FIGURE 7.4 Mass fraction of sol as a function of the mole ratio of amine and epoxy groups; (*) experimental data, (—) calculated dependences taking epoxy conversion, x, into account. s ¼ PGE PGE þ 2 DGEBA and r ¼ 4 MDA PGE þ 2 DGEBA where PGE ¼ phenyl glycidyl ether (monoepoxy), DGEBA ¼ diglycidyl ether of bisphenol A (diepoxy), and MDA ¼ methylene dianiline. (a) For all the series with constant s, critical molar ratios for gelation (r c ), were determined. The gel point was characterized by the first occurrence of an insoluble gel in dimethyl formamide, DMF. (b) The residual concentration of unreacted epoxy groups was determined by infrared spectroscopy and used to determine the final conversion, x. (c) The mass fraction of sol, W sol , was determined by multiple extraction of networks in DMF, at 1308C. (Ilavsky et al., 1984 – Copyright 2001 – Reprinted by permission of John Wiley & Sons, Inc.) takes place (Wu and Bauer, 1985). This process is driven by thermo- dynamics (change of the interaction energies along polymerization) and leads to inhomogeneous networks. b. Inhomogeneities Induced by Chemical Reaction Inhomogeneities can also be formed during reaction without any help of physical interaction; i.e., just by chemistry. For silane condensation reac- tions in solution (the so-called sol–gel chemistry) but also in bulk, the experimental gel point conversions are very much higher than the theoretical values due to the formation of cagelike structures (Chapter 3). In some cases, using trifunctional silanes with large organic substituents, it is possible to obtain only cage structures without gelation (Fasce et al., 1999). Figure 7.5 shows typical cagelike structures. Such a high percentage of intramole- cular reactions is very specific to silane condensation and quite unusual for most step polymerizations in bulk. Inhomogeneities can be formed in step-polymerization processes when more than two monomers are reacted together. Figure 7.6a represents a crosslinked mixture of a long diol (polyether or polyester, M n $ 1000– 5000 g mol À1 ), a low-molar-mass triol (trimethylol propane, TMP), and a diisocyanate. Chemical or topological clusters are assemblies of covalently bonded units of one kind differing in some properties from the surrounding matrix. The polymer network consists of a dispersion of hard clusters in a soft matrix (Nabeth et al., 1996). The presence of these inhomogeneities can be clearly recorded by SAXS. In this case of three-monomer polyurethane synthesis, there is no thermodynamic driving force for phase separation. The formation of clus- ters is fully controlled by the initial composition of the system, the reactivity of functional groups, and the network formation history (one or two stages, macrodiol or triol reacted with diisocyanate first, etc.). The presence of hard clusters affects mechanical properties. The major problem is the way to define elastically active network chains (EANC) and crosslinks (Chapter 3, Fig. 3.3). It has been demonstrated that hard clusters must be considered as multifunctional crosslinks (f c ¼ 6 in Fig. 7.6a) while macrodiol chains behave as EANC. Chemical clusters can be obtained also with two monomers, when two reaction mechanisms are in competition, favoring formation of regions of higher and lower crosslink densities. This situation is more complex and more difficult to control. It is certainly the case for dicyanodiamide (Dicy)-cured epoxies: with this hardener an accelerator is always used and a competition between step (epoxy–amine addition) and chain (epoxy homopolymerization) occurs (Chapter 2), leading to inhomogeneous networks. Are Cured Thermosets Inhomogeneous? 213 214 Chapter 7 FIGURE 7.5 (a) Schematic representation of an octosilsesquinoxane, (RSiO 1.5 ) 8 ; (b) typical precursor of a SSQQ; (c) incompletely condensed ‘‘T8 (OH) 2 ’’ isomers. (Reprinted with permission from Fasce et al., 1999. Copyright 2001. American Chemical Society) c. Both Chemical Clustering and Microphase Segregation In the case of Fig. 7.6a the cluster formation and the size distribution can be influenced not only by chemical reactions but also by partial miscibility of the substructures during reaction. Polyurethane networks prepared from polyolefin instead of polyester or polyether as macrodiol, can serve as an example. In this particular case an agglomeration of hard domains takes place in the pregel stage, produced by a thermodynamic driving force. The ‘‘sol–gel’’ chemistry has also been used to prepare inorganic inho- mogeneities in an organic matrix. Silane end-capped macrodiols can be used. Hydrolysis and condensation of alkoxy silane groups lead to inorganic hard clusters (Fig. 7.6b). Intramolecular reactions and the miscibility of the soft-segment chains with the relatively polar crosslinks determine the size distribution of the clusters (nanofillers). Are Cured Thermosets Inhomogeneous? 215 FIGURE 7.6 Schematic representation of hard clusters in: (a) a polyurethane network composed of a long diol, a triol, and a diisocyanate (polyaddition reactions, Chapter 2). (b) a hybrid inorganic–organic network composed of a silane end-capped long diol (polycondensation reactions, Chapter 2). [...]... be made 224 FIGURE 7. 11 Chapter 7 Different types of preformed particles REFERENCES Andrady AL, Sefcik MD, J Polym Sci Phys., 21, 2453–2463 (1983) ´ Barbeau P, Gerard JF, Magny B, Pascault JP, Vigier G, J Polym Sci Phys., 37, 919–9 37 (1999) Boots HMJ, Physica 147A, 90–98 (19 87) ´ Cuney S, Gerard JF, Dumon M, Pascault JP, Vigier G and Dusek K, J Appl Polym Sci., 65, 2 373 –2386 (19 97) Dusek K, In Polymer... build nanosized ordered regions in the networks Different architectures, such as block copolymers, crosslinked microparticles, hyperbranched polymers and dendrimers, have emerged (Fig 7. 11) Crosslinked microparticles (‘‘microgels’’) can be described as polymer particles with sizes in the submicrometer range and with particular characteristics, such as permanent shape, surface area, and solubility The... (eds), Plenum Press, New York, 1 971 , pp 245–260 Dusek K, In Epoxy Resins and Composites IV, Dusek K (ed.), Adv Polym Sci., 80, 173 –188 (1986) Dusek K, Collect Czech Chem Commun., 58, 2245–2265 (1993) Are Cured Thermosets Inhomogeneous? 225 Dusek K, Bleha M, Lunak S, J Polym Sci Chem., 15, 2393–2402 (1 977 ) Dusek K, Plestil J, Lednicky F, Lunak S, Polymer 19, 393–3 97 (1 978 ) ´ Fasce DP, Williams RJJ, Mechin... Llauro MF, Petiaud R, Macromolecules, 32, 475 7– 476 3 (1999) Funke W, Okay O, Muller BJ, Adv Polym Sci., 136, 139–234 (1998) ¨ Gabrys B, Tomlins PE In Encyclopedia of Polymer Science and Engineering, 2nd ` edn, Mark HF, Bikales NM, Overberger CG, Menges G (eds), Scattering, 15, 1–68 (1989) Hsu CP, Lee LJ, Polymer, 34, No 21 Part 1, 4496–2505; Part 2, 4506–4515; and Part 3, 4516–4523 (1993) Ilavsky M, Bogdanova...216 Chapter 7 It has been demonstrated by SAXS analysis (Fig 7. 7, Cuney et al., 19 97) , that the control of the intermolecular reactions by chemical reactivities (polycaprolactone, PCL20 case) only gives loose, more or less randomly branched structures of fractal types, whereas thermodynamic microphase segregation (hydrogenated polybutadiene, HPBD case) yields compact particles, possibly with... was cast into a mold and cured for 24 h at 1008C under pressure, and then postcured at 1508C for 12 h (Cuney et al., 19 97 – Copyright 2001, Reprinted by permission of John Wiley & Sons, Inc.) Are Cured Thermosets Inhomogeneous? 2 17 7.3 THERMOSETS FROM CHAINPOLYMERIZATION MECHANISM 7. 3.1 Chainwise Polymerization is Intrinsically an Inhomogeneous Process As explained in Chapters 2 and 3, the step and... largest extent of inhomogeneity (Fig 7. 8a) The initiation rate was artificially increased by limiting the chain length to at most 20 steps (Fig 7. 8b) and 2 steps (Fig 7. 8c) There is a continuous increase in the uniformity of the polymer distribution in space with the increase in the ratio between initiation and propagation rates For the limiting case plotted in Fig 7. 8c, the chain process has been practically... 265– 278 (1984) Labana SS, Newman S and Chompff AJ In Polymer Networks, Chompff AJ, Newman S (eds), Plenum Press, New York 1 971 , pp 453–4 87 Nabeth B, Pascault JP, Dusek K, J Polym Sci., Polym Phys Ed., 34, 1031–1054 (1996) Newkome GR, Moorefield CN, Vogtle P, Dendritic Molecules: Concepts, Syntheses, Perspectives, Wiley-VCH, Weinheim, 1996 Racich JL, Koutsky JA, J Appl Polym Sci., 20, 2111–2129 (1 976 )... expected to persist in the final structure (qualitatively represented by Fig 7. 1) (d) Thermodynamic effects may also be present For example, if CMP are not soluble in the mixture of monomers, they will agglomerate (phaseseparate) in the first stages of the polymerization, leading to a broad distribution of particle sizes (including large particles) A well-known example is the cure of unsaturated polyesters,... step polymerization after a conversion of 25% of the reactive groups (Reprinted from Boots, 19 87, Copyright 2001 with permission from Elsevier) 7. 3.2 Network Formation in the Case of Free-Radical Polymerization a Spatial Correlation and Apparent Reactivity of Double Bonds As described qualitatively in Fig 7. 8a, the fast chain propagation with respect to initiation is both a source of inhomogeneities . block copolymers, crosslinked micro- particles, hyperbranched polymers and dendrimers, have emerged (Fig. 7. 11). Crosslinked microparticles (‘‘microgels’’) can be described as poly- mer particles. Sci. Phys., 37, 919–9 37 (1999). Boots HMJ, Physica 147A, 90–98 (19 87) . Cuney S, Ge ´ rard JF, Dumon M, Pascault JP, Vigier G and Dusek K, J. Appl. Polym. Sci., 65, 2 373 –2386 (19 97) . Dusek K,. 7 FIGURE 7. 11 Different types of preformed particles. Dusek K, Bleha M, Lunak S, J. Polym. Sci. Chem., 15, 2393–2402 (1 977 ). Dusek K, Plestil J, Lednicky F, Lunak S, Polymer 19, 393–3 97 (1 978 ). Fasce