February 2003 D1. Modeling of Polymer Phase Transitions This project will: investigate dynamic models for phase transitions from liquid to other states, such as glass, gel, and crystal for polymers, polymer mixtures, and colloids. The models, based on population balances (distribution kinetics), will be applied to spinodal decomposition, gelation, nucleation, growth and ripening, and glass transition. Primary Faculty co-Advisors Benjamin J. McCoy, Chemical Engineering, (polymer reaction kinetics, phase transition dynamics, and complex phenomena) Gudrun Schmidt, Chemistry, (nanotechnology, complex molecular self-assemblies, smart fluid applications) Off-campus Participant: George Roberts, Department of Chemical Engineering, North Carolina State University. Professor Roberts has published extensively on polymerization kinetics, particularly for reactions in environmentally safe solvents, such as carbon dioxide. Technical Proposal: Macromolecular Phase Transition Dynamics Background and Introduction The dynamics of phase transitions have been a topic of intensive research, and many of the governing principles are fairly well understood. Comprehensive and systematic approaches are still needed, however, to unify and systematize the understanding of its many facets. A broad area of study, phase transition dynamics embraces first-order transitions, including nucleation, growth and coarsening of particle (liquid-solid) or droplet (vapor-liquid) size distributions. A variation on this process is spinodal decomposition, or spontaneous and barrierless condensation. A transition that is not first-order is the glass transition from liquid to solid glass, characterized by the absence of nucleation. The gel transition, occurring for example during aggregation and intermolecular crosslinking of macromolecules, is another process that is of current active investigation because of its importance in polymer processing. These phenomena are all characterized by size-distributed heterogeneities, usually manifested as different sized clusters or macromolecules. The kinetics and dynamics of size distributions, whether for particulates or macromolecules, can be mathematically described by population balance modeling. Population balance models represent a powerful approach to quantifying complex chemical and engineering phenomena. The proposed research concerns the application of this technique to phase transition kinetics and dynamics involving polymers, polymer mixtures, and colloids. Phase Transition Dynamics The three steps of a first-order phase transition are nucleation (either homogeneous or heterogeneous), growth, and Ostwald ripening (coarsening). In spinodal decomposition, the phase change is spontaneous, with no energy barrier for nucleation. 1 February 2003 Homogeneous nucleation can be described via classical nucleation theory, such that volume and surface energies combine to form a free energy barrier that must be overcome to form a nucleus of critical size. The nucleation step is difficult to observe experimentally, due to the very small size of the critical nuclei, but it dictates the rest of the growth process. For polydisperse colloid suspensions, simulations have shown that the free energy barrier may increase, after going through a minimum, as the density increases [1]. Thus, a slowdown in nucleation rate at high density may be related to an increase in the free energy barrier, not a slow-down in the kinetics as fluid motion becomes sluggish. This same simulation relates the probability of formation of a critical nucleus to the polydispersity of the suspension, such that polydisperse suspension may not crystallize at all, even when compressed [2]. According to classical nucleation theory, the free energy barrier variation with polydispersity is due to an increase of the interfacial free energy, (if the chemical potential difference between the solid and liquid states is assumed constant). Turnbull has suggested that the interfacial free energy should be proportional to the latent heat of fusion, but the simulation indicates that when the latent heat becomes zero, the interfacial free energy, and thus the free energy barrier, are non-zero [2]. In the case of zeolite A crystal growth at room temperature, an amorphous gel formed prior to nucleation occurring [3]. This type of behavior is also not explained via classical nucleation theory. So, the classical theory of homogeneous nucleation does not appear to account for all observed or modeled behaviors, and thus another approach to explaining this critical step in phase transition is needed. Various mechanisms by which growth of a nucleus may occur have been proposed, which range from solution-mediated transport to hydrogel transformation processes [3]. The zeolite A experiments mentioned above reported that crystals grew at the expense of the amorphous agglomerates, resulting in crystals that were similar in size to the amorphous precursor particles. When the solution was heated above room temperature, however, much larger crystals resulted, presumably assembled from material drawn from solution [3]. So, two different growth mechanisms are seen in the same solution, depending on the temperature at which the crystallization is carried out. These two mechanisms result in crystals of different sizes and different properties. Thus, a single explanation for growth that applies to all phase transitions is unlikely to be correct, and a theory that allows for different mechanisms in competition with one another is desirable. Coarsening or ripening, the last step in a first-order transition, involves the dissolution of small crystals with their mass going to larger crystals, which continue to grow. In essence, the small crystals give up their molecules so that the larger crystals increase in mass. The theoretical basis for this behavior is the Gibbs-Thomson relation, in which smaller particles have a relatively larger interfacial energy and are thus more soluble than larger particles. Once the smaller particles shrink to the critical nucleus size, they become thermodynamically unstable and disintegrate [4 & 5]. Conventional ripening theories, such as the Lifshitz, Slyozov, Wagner (LSW) theory and modifications [6] account for denucleation indirectly, through the mass balance. Although the LSW theory is the foundation for understanding Ostwald ripening, the form of the particle size distribution predicted by LSW theory does not match experimental results in several cases [6]. The distribution kinetics theory [6], based on a population balance approach, 2 February 2003 improved on the LSW model by directly accounting for denucleation and interactions between particles, producing a model that is valid at all times from zero to infinity. Another type of phase transition is the glass transition, in which a metastable liquid becomes a highly viscous, molecularly disordered glass when cooled or compressed rapidly, such that nucleation and crystallization do not occur. This type of phase transition is also seen in colloidal suspensions. Clusters (i.e. heterogeneous structures) are frequently used to explain this phenomenon, and competition between disordered clustering and ordered crystallization can lead to glass formation. Experimental observation of clusters in supercooled colloidal fluids has been reported [17]. Thus, a model that allows for competing cluster processes is clearly applicable. The distribution kinetics theory has been successfully used to model the viscosity and dielectric relaxation time dependence on temperature for glassformers [15]. Gelation represents an irreversible phase transition, in that once the gel state is reached, the polymer will not return to the sol state without a chemical change in the solution, such as a pH modification. Flory developed statistical models to explain how polymer gelation occurs through cross-linking. A novel distribution kinetics approach, in which two distribution variables are considered simultaneously, molecular weight and number of cross-links, yields a general model for gelation [9 & 10]. Cross-linking also occurs in polymer-clay nanocomposites [18], and gels of these compounds are of interest for their unique properties. In polymer-clay composites, bonds exist between polymer and clay, as well as between polymer molecules. The properties of polymer-clay nanocomposites can be influenced by application of a shearing force, because the bonds between clay and polymer are weaker than bonds between polymers. The colloidal and rheological properties of polymer-clay gels can potentially be modeled by a population balance approach similar to that applied to gelation. The process of spinodal decomposition, a non-first order phase transition, has many applications in materials science and engineering, including polymer blending. For some spinodal decompositions, the initial segregation phase of the decomposition is fast, on the order of a few milliseconds. For two liquid phases separating from one miscible liquid phase, experiments have demonstrated that following the initial separation, small microdomains grow by diffusion and coalescence [7]. Experiments in which the phase transition is retarded by considering only very shallow quenches below the critical temperature demonstrate that these domains are at local equilibrium with each other, and they grow first by diffusion, and then by convection [7]. Similar results are obtained in very viscous polymer blends [8]. Simulations indicate, however, that deeply quenched liquid mixtures are never at local equilibrium, because segregation and growth occur simultaneously [7]. The last stage of a spinodal decomposition phase transition for a liquid mixture involves coarsening of the phase-separated droplets. During this last stage, once the droplets are large enough that buoyancy dominates surface tension effects, the mixture separates by gravity. Thus the mechanisms that dominate a spinodal decomposition phase transition depend on the process. The process and the competing mechanisms underlying the phase transition should both be considered in a mathematical model simulation. 3 February 2003 Proposed Research In this work, we propose to model the dynamics of polymer phase transitions, including spinodal decomposition, glass transition, nucleation, growth and ripening, and gelation. Fluids consisting of a single polymer, a mixture of polymers, and colloids composed of polymer will be considered. The modeling technique to be applied is distribution kinetics, or population balance modeling. In general terms, phase transitions can be regarded as reaction-like mechanisms in which molecules cluster together, or break apart, at various rates, resulting in a distribution of cluster sizes that changes with time until equilibrium (or steady state) is reached. The rates of these reaction-like processes may depend on such variables as cluster size, temperature, and pressure. Population balance equations can be written for the clusters as well as for monomer, and mathematically transformed to moment equations, allowing calculation of moments representing molar concentration, number average molecular weight, and polydispersity. The various moments can be related to properties of the fluid, such as viscosity in the case of the glass transition, or supersaturation in the case of nucleation and growth. For gelation by aggregation accompanied by inter- or intramolecular crosslinking, the number of crosslinks c-an also be incorporated into the population balance equation [9 & 10]. Previous work in population balance modeling has demonstrated that distribution kinetics for nucleation, reversible growth, and reversible aggregation processes can be used to model such phase transition behaviors as cluster growth by monomer addition, cluster dissolution by monomer dissociation, classical nucleation by monomer addition, denucleation by monomer dissociation, ripening, vitrification, and induction time for precipitation due to low nucleation rate [11]. This earlier work, which assumed that the associated rates for each process were functions only of temperature, has been extended for the metastable phase transition case to include rates dependent on the cluster mass [12 & 13]. Further extensions accounted for the Gibbs-Thomson effects on equilibrium solubility, phase transition energies, interfacial energies, and the critical nucleus size for denucleation [14]. For the glass transition, only temperature dependence has been modeled to date [15]. For a first-order condensation phase transition, nucleation may occur via a homogenous route, in which a free energy barrier must be surmounted to form a critical nucleus, or a heterogeneous route, in which nucleation sites allow monomer deposition. Formation of critical nuclei can be viewed as a series of monomer-addition reactions. For the homogeneous route, the rate of nucleation depends on the size of the free energy barrier. This free energy barrier can be related to the supersaturation, which is in turn given by m(0)/m(0)eq, the molar concentration of monomer/molar concentration of monomer at equilibrium. For heterogeneous nucleation, the same reaction scheme can be assumed to form the nuclei, but without a free energy barrier-dependent rate. Once the nuclei are formed, they grow via monomer addition. Growth via monomer addition can be modeled similarly to the nucleation scheme above, using a “reaction” of the form, M(xm) + C(x) C(x + xm), where M is monomer, C represents clusters, and x represents size of the cluster. Since this process is random, the resulting clusters vary in size, yielding a cluster size 4 February 2003 distribution. Clusters may also coalesce and break up, which can be described via a process of the type, C(x') + C(x) C(x + x'). In the ripening stage, due to the reduction in interfacial area as clusters increase in size, large clusters continue to grow and small clusters dissolve. This process can be modeled similarly to the growth stage by introducing the Gibbs-Thomson relations. Population balances on monomer, M, and clusters, C, can then be written, in terms of the rates, as well as any source/sink terms in the balance. For example, cluster denucleation would appear as a sink in the cluster balance, but would be a source term in the monomer balance, since monomers result when the clusters denucleate. Given a model of this form, expressions for the reaction rates must be determined, and the results tested against experimental data for polymer condensation. Recent results [4, 5, 6, 12, 14] need to be extended to polymers. An additional validation of the model is the comparison with molecular-level models of phase transitions, such as lattice-Boltzmann models [16]. In addition, our model has not yet been applied to polymer mixtures that phase separate either below the upper critical solution temperature or above the lower critical solution temperature. For a mixture, the phase separation occurs via the same processes as described above, so that a similar model may be applicable. Recent experiments describe a distribution of droplets of varying sizes that grow with time [8], indicating that distribution kinetics has potential to accurately model this phenomenon. As polymer colloids have a particle size distribution, a population balance model may also prove useful in describing the properties of such a solution, as well as its ability to crystallize. In the case of spinodal decomposition, there is no barrier to the phase change, so that it occurs spontaneously. Thus, a reaction model for a condensation would involve the same equations as above for growth and monomer addition. Such a model could be applied to the spontaneous separation into two phases of formerly miscible polymers upon lowering of temperature below the upper critical solution temperature, as well as the condensation of a single polymer upon lowering of pressure. A distribution kinetics approach appears valid for these cases, because light-scattering experiments show clusters that increase in size with time during spinodal decomposition. We will explore the pressure-induced solidification of a metastable liquid or colloidal solution (glass transition), for example, polymer solution or melt. A previous model [15] for cluster kinetics of glass formation will be extended, via transition state theory, to include activation volumes in the rate coefficients for monomer-cluster additiondissociation and cluster agglomeration-breakage. The glass formation kinetics are based on the dynamics of hypothesized cluster-monomer distributions in liquids. The heterogeneities in glassy materials have been well documented [17]. The consequences of this hypothesis for metastable liquids that do not nucleate are interpreted in terms of fragility plots for the pressure effect on viscosity and dielectric relaxation time. The proposed model yields entropy relations that afford an explanation for the effect of pressure on entropy, and provides free volume ratios that allow representation of the viscosity. Given values of pressure at the glass transition and when the material is fluid, a single volume-difference parameter is needed to complete the correlation. 5 February 2003 Previous work on gelation models [9 & 10] will be extended to polymer-clay gels and solutions, in which bonds between polymer and clay can be broken via application of a shearing force [18]. In the distribution kinetics model, cross-linking has been described by the following “reactions”: Pi(x) + Pj(x') → Pi+j+1(x + x'), Where Pi(x) is a macromolecule with molecular weight x and i crosslinks. The above equation describes the loss of Pi(x), whereas the gain of Pi(x) is given by: Pi-j-1(x-x') + Pj(x') → Pi(x). For the polymer-clay nanocomposite, in which polymer is bonded to clay and to other polymer, the above reaction scheme would have to be expanded to include clay, and the clay reactions must be reversible, since bonds between polymer and clay can be broken. Once a model of the polymer-clay composite has been developed, it could be used to explain the response of this substance to shear, as this response is believed to result from changes in polymer-clay interactions [18]. References 1. Oxtoby, D.W. Diversity suppresses growth. Nature 2001, 413, 694 and references therein. 2. Auer, S.; Frenkel, D. Suppression of crystal nucleation in polydisperse colloids due to increase of the surface free energy. Nature 2001, 413, 711 3. Mintova, S.; Olson, N.H.; Valtchev, V.; Bein, T. Mechanism of Zeolite A Nanocrystal Growth from Colloids at Room Temperature. Science 1999, 283, 958. 4. Madras, G.; McCoy, B.J. Transition from nucleation and growth to Ostwald ripening. Chemical Engineering Science 2002, 57, 3809. 5. Madras, G.; McCoy, B.J. Denucleation Rates during Ostwald ripening: Distribution kinetics of unstable clusters. Journal of Chemical Physics 2002, 117, 6607. 6. Madras, G.; McCoy, B.J. Continuous distribution theory for Ostwald ripening: comparison with the LSW approach. Chemical Engineering Science 2003, 58, 2903-2909 and references therein. 7. Gupta, R.; Mauri, R.; Shinnar, R. Phase Separation of Liquid Mixtures in the Presence of Surfactants. Industrial Engineering Chemistry and Research 1999, 38, 2418-2424 and references therein. 8. Califano, F.; Mauri, R. Drop Size Evolution during the Phase Separation of Liquid Mixtures. Industrial Engineering Chemistry and Research 2004, 43, 349353 and references therein. 9. Li, R.J.; McCoy, B.J. Crosslinking kinetics: Partitioning according to number of crosslinks. Macromolecular Theory and Simulations 2004, 13, 203-218. 10. Li, R.J.; McCoy, B.J. Inter- and Intramolecular Crosslinking Kinetics: Partitioning According to Number of Crosslinks. Macromolecular Rapid Communications 2004, 25, 1059-1063. 6 February 2003 11. McCoy, B.J. Distribution kinetics modeling of nucleation, growth, and aggregation processes. Industrial & Engineering Chemistry Research 2001, 40, 5147-5154. 12. Madras, G.; McCoy, B.J. Ostwald ripening with size-dependent rates: Similarity and power-law solutions. Journal of Chemical Physics 2002, 117, 8042-8049. 13. Madras, G.; McCoy, B.J. Dynamics of crystal size distributions with sizedependent rates. Journal of Crystal Growth 2002, 243, 204-213. 14. Madras, G.; McCoy, B.J. Growth and ripening kinetics of crystalline polymorphs. Crystal Growth & Design 2003, 3, 981-990. 15. McCoy, B.J. Cluster Kinetics for Glassforming Materials. Journal of Physics and Chemistry of Solids 2002, 63, 1967-1974. 16. H. Başağaoğlu, C.T. Green, P. Meakin, B.J. McCoy, Lattice-Boltzmann Simulation of 2D Phase Transition with Ostwald Ripening, J. Chem. Phys. (2004); submitted. 17. Ediger, M.D. Spatially Heterogenous Dunamics in Supercooled liquids and the glass transition, Annual Review of Physical Chemistry 2000, 51, 99. 18. Malwitz, M.M.; Butler, P.D.; Porcar, L.; Angelette, D.P.; Schmidt, G. Orientation and Relaxation of Polymer-Clay Solutions Studied by Rheology and Small-Angle Neutron Scattering. Journal of Polymer Science: Part B: Polymer Physics 2004, 42, in press. Number of IGERT apprentices to be recruited and probable home departments: 2 total 1, chemical engineering 1, chemistry Consistency with the Macromolecular Education, Research & Training theme: This project requires the students to understand polymers, and the methods used to characterize and model the dynamics of their phase transitions. The education provided via the courses in macromolecules will provide the foundation in polymers to pursue the project, and the research will provide training in polymer modeling both at the macroscopic and microscopic level. How does the project form a vector cross-product of existing research themes by the participants? Existing research directions. Dr. McCoy’s current research interests related to this project include phase transition dynamics and distribution kinetics. He and collaborators have published numerous papers applying these modeling techniques to polymers, such as the kinetics of polymers in solution, polymer decomposition/degradation, polymerization reactions, and thermogravimetric analysis of polymers. He is a full professor in chemical engineering, holding the Gordon A. & Mary Cain endowed chair. Dr. Schmidt’s current research interests include the synthesis, development, and characterization of new multi-component materials with hybrid properties relevant to polymer and colloid science. In particular, her work on polymer-clay nanocomposites 7 February 2003 interfaces well with Dr. McCoy’s modeling expertise, in that the data generated by her group could be used to validate models developed by Dr. McCoy’s group. Dr. Schmidt is an assistant professor in the chemistry department. New research direction. This research project directly joins the research areas of the two professors involved, by attempting to validate models developed by Dr. McCoy’s group of novel polymer-clay composites produced and characterized by Dr. Schmidt’s group. A synergistic collaboration between the two groups has the potential to yield interesting new results. How do students benefit from the team-oriented research, beyond what would be available to them from either advisor separately? The inter-disciplinary nature of this project allows the chemical engineering student exposure to synthesis, development, and characterization of nanocomposite materials, an area not normally the province of engineers, while exposing the chemistry student to distribution-level models, a technique not normally seen in chemistry. Briefly describe the support level available to each individual faculty or off-campus participant (i.e., without IGERT) Dr. McCoy is independently supported for research in related fields by the Gordon A. and Mary Cain endowed Chair. Dr. Schmidt is the recipient of an NSF Career Award, and thus is also independently supported for research in related fields. Interdisciplinary strengths of the team project: Dr. McCoy has a broad and extensive background in mathematical modeling applied to a variety of areas in chemical engineering. Dr. Schmidt has been working on polymerclay nanocomposites for several years, and is now progressing to detailed characterization of these materials. By joining their efforts, these novel polymer-clay materials can be modeled, and the models validated against experimental data, something that neither could do alone. Commitment of faculty & off-campus participants to work side-by-side with apprentices: Professor McCoy, a full professor in the department of chemical engineering, promises daily interaction with the student during the approximately one month period of the apprenticeship. Consultations with Professor Schmidt, an assistant professor in the chemistry department, and her group will be frequent. 8