UC Berkeley

Polymer composites combine the physical properties of two or more distinct chemical species into a single material. Blends of two homopolymers, mixed homogeneously or structured on the nanoscale, are particularly useful nanocomposite materials because ordering occurs via thermodynamic equilibration rather than costly nanofabrication. However, the majority of homopolymer pairs are immiscible, leading to mechanically unstable materials. This work explores one route for compatibilizing two immiscible homopolymers (components A and B) via the addition of a tailored diblock copolymer surfactant (component A-C). The A-block of the copolymer was selected on the basis of having a neutral interaction with the A homopolymer, and the C-block had a favorable interaction with the B homopolymer. The favorable interaction between species B and C was first examined by preparing binary blends of B and C homopolymers. A detailed thermodynamic study explored the effects of blend composition and homopolymer chain length on the thermodynamic phase behavior of binary B/C blends. The results were used to design A/B/A-C ternary blends where the favorable interaction between homopolymer B and the C-block enabled the copolymer surfactant to efficiently stabilize the interface between nanoscale-ordered domains of A and B. The phase behavior of the resulting polymer nanocomposites was studied as a function of the composition of the blend and the chain lengths of the A and B homopolymers. These studies provide new insight into the tunability of of polymer nanocomposite materials by controlling parameters that have not been studied previously.

Polymer pairs that are miscible over a large range of chain lengths, N,_B and N_C, are interesting to thermodynamic studies because the Flory-Huggins interaction parameter, χ, can be measured using scattering techniques and the theoretical framework of the Random Phase Approximation. Most tabulated χ values have been measured over a limited range of chain lengths small enough for the polymers to homogenize; such studies are largely constrained to the vicinity of N,_B = N,_C. This dissertation presents the most comprehensive study of χ to date for a single pair of homopolymers.

Polyisobutylene (component B) and deuterated polybutadiene with 63 % 1,2 addition (component C) were selected for this study because they exhibit a large window of miscibility and may be tailored to cross the spinodal at experimentally accessible temperatures. Binary blends were designed across a range of values for N,_B/N,_C and the composition of the blend, φ_B, to study the effect of these parameters on the measured value, χ_sc. In addition to the strict temperature dependence presumed for χ, this study documented a composition and molecular weight dependence. The empirical expression for χsc, measured using small angle neutron scattering, was three times more dependent on composition then the expression for χ used to predict thermodynamic behavior. Despite this three-fold diminished dependence on φ_B, the composition-dependent χ profoundly affected the phase behavior of binary blends.

Binary B/C blends exhibited macrophase separation upon heating (above a threshold chain length), enabling experimental determination of the binodal and spinodal. These measured quantities were compared to predictions using Flory-Huggins Theory with the composition- and molecular weight-dependent χ. Phase diagrams are expected to be symmetric in the vicinity of N,_B = N,_C, with the critical point located at φ_B,crit = 0.5. However, both the measured and predicted phase diagrams were asymmetric in the vicinity of N,_B = N,_C, and increasingly symmetric as the value of N,_B/N,_C was decreased. A range of values was studied for N,_B/N,_C = 1, and in all cases φ_B,crit was found to be < 0.5, in stark contrast to the expectation of Flory-Huggins Theory that φ_B,crit = 0.5. This effect was shown to result from the combined effects of a composition-dependent χ and N,_B/N,_C removed from values of 1. Remarkable agreement was obtained between the predicted phase diagrams and measured phase transitions, over a range of values for N,_B/N,_C and φ_B, by accounting for the composition and molecular weight dependence of χ.

The miscibility of binary B/C blends was used as the basis for designing a diblock copolymer (component A-C) to order immiscible binary blends of polyisobutylene and deuterated polybutadiene with 89 % 1,2 addition (component A). The copolymer comprised one block chemically identical to component C (miscible in component B) and one block chemically identical to component A. This is in contrast to the majority of ternary blend studies which comprise A/B/A-B polymer systems with neutral interactions between each homopolymer and the corresponding block of the diblock copolymer. Ternary A/B/A-C blends exhibit a favorable interaction between the B homopolymer and C block, demonstrated by the miscibility of B/C blends. The A-C diblock copolymer surfactant can produce microstructures when added to A/B blends at much lower concentrations of copolymer than for an analagous A-B copolymer. Previous studies have only considered the case N_A = N_B and a symmetric diblock copolymer (N_A-block = N_C-block). In the present work, symmetric diblock copolymers were added to critical A/B blends. The values of N_A/N_B were varied over two orders of magnitude. For each blend, the ratio of A:B was fixed by the Flory-Huggins Theory prediction for the critical point (which depends only on N_A/N_B), and a constant amount of copolymer was used for all blends. By creating blends with a wide range of values for N_A/N_B, the study accessed critical compositions, φ_A,crit, well removed from the typical value of 0.5 (on a copolymer-free basis). The resulting phase behavior correlated directly with N_A/N_B, suggesting that the microstructures observed in a blend could be tuned not only by the conventional method of changing the amount of copolymer, but also by adjusting the ratio N_A/N_B.

Lamellar or microemulsion phases were observed over a temperature window for nearly all of the A/B/A-C blends studied. The former represent an ordered microphase and the latter a disordered microphase, each with unique applications. Previous work has attempted to distinguish the scattering profiles of lamellar phases from those of microemulsions by fitting both with the Teubner-Strey equation for microemulsions. The lamellar phase was judged to exist when the microemulsion fit did not capture the entire range of the scattering profile, otherwise the phase was presumed to be a microemulsion. This dissertation introduces the use of lamellar structure factor that fits scattering profiles unsuitable for the microemulsion fit. In addition, the lamellar fits include as adjustable parameters the size of each microdomain and corresponding interfacial width. These fit values agree quantitatively with independently generated predictions using self-consistent field theory, indicating a broad understanding of the physical parameters that affect thermodynamic behavior in the A/B/A-C system studied.

The composition of a blend, in particular the concentration of diblock copolymer, is known to affect the phase behavior, however most studies have only considered blends where equal amounts of A and B are blended with copolymer and φ_A,crit = 0.5 (on a copolymer-free basis). This dissertation presents a study for which the concentration of diblock copolymer was fixed and the composition of the A and B homopolymers was systematically varied across a range of compositions including φ_A,crit. The experiment corresponded to tracing the copolymer isopleth on a ternary phase prism. Theoretical groups have predicted a rich phase behavior along the isopleth for similar ternary systems, however, the observed phase behavior was quantitatively identical for all blends studied. Self-consistent field theory predictions agreed with fit values of the domain spacing and microdomain widths. There was no discernible correlation between φ_A and phase behavior. This finding, and that of the study with critical A/B/A-C blends, together suggest that N_A/N_B correlates strongly with the phase behavior of a blend, while φ_A does not. This relationship, captured by mean-field theory, provides a simple method for tuning the phase behavior of polymer nanocomposites.