The proton-exchange membrane (PEM) is a central, and often performance-limiting, component of all-solid H2/O2fuel cells. Nafion®, the most widely used PEM, consists of a perfluorinated polymer that combines a hydrophobic Teflon-like backbone with hydrophilic ionic side groups. It stands out among polymer materials for its high, selective permeability to water and small cations. Generous funding of our research on Nafion has been provided by the Department of Energy, Basic Energy Sciences via the Ames Laboratory.
High-resolution 13C NMR of fluoropolymers. In spite of the technological significance of perfluorinated polymers such as Nafion, or poly(tetrafluoroethylene), PTFE/Teflon®, no 13C solid-state nuclear magnetic resonance (NMR) spectra of these temperature- and solvent-resistant materials could be found in the literature prior to 2001. The standard 13C spectra of Nafion and Teflon are broadened due to the large 19F chemical-shift anisotropy, which prevents on-resonance 19F decoupling. We have obtained the first high-resolution 13C NMR spectra of solid perfluorinated polymers by combining 28-kHz magic-angle spinning (MAS) with rotation-synchronized 19F 180°-pulses. The small line width shows that most Nafion backbone segments are helical and conformationally ordered, even though Nafion is a random copolymer. Conformational disorder is concentrated at the branch points. [Macromolecules 37, 5995-6003 (2004)] Furthermore, motional narrowing of 13C-19F dipolar splittings proved that most chain segments between branch points rotate by more than 150o around their helix axes. This rigidity of the backbone excludes many models of Nafion that are based on the assumption of random coiling. [Macromol. Chem. Phys. 208, 2189-2203 (2007)] Nevertheless, the helices do not pack into well-ordered bundles, according to orientational correlation data from 19F CODEX (centerband-only detection of exchange) NMR.
Quantitative SAXS simulations of the nanostructure of Nafion. Important aspects of the long elusive nanometer-scale structure that underlies many of these outstanding properties of Nafion have now been conclusively determined by a quantitative analysis of small-angle scattering data, using a novel approach based on numerical Fourier transformation (see above). The characteristic "ionomer peak", see inset in Figure 2d, arises from long parallel but otherwise randomly packed water channels surrounded by partially hydrophilic sidebranches, forming inverted-micelle cylinders. They are stabilized by the rigid polymer backbones proven by NMR. At 20 vol% water, the water channels have diameters between 1.8 and 3.5 nm, with a 2.4-nm average. Nafion crystallites (~10 vol%), which form physical crosslinks crucial for the mechanical properties of Nafion films, are elongated and parallel to the water channels, with cross sections of ~(5 nm)2. Simulations for various other models of Nafion, including Gierke's cluster and the polymer-bundle model, do not match the scattering data. The new model can explain important features of Nafion, including fast diffusion of water and protons through Nafion and its persistence at low temperatures. It also provides a valid target for the design of other, cheaper ionic polymers that could replace Nafion. [Nature Mater. 7, 75-83 (2008)]
Figure 2. Parallel water-channel (inverted-micelle cylinder) model of Nafion. a, Two views of an inverted micelle cylinder, with the polymer backbones on the outside and the ionic sidegroups lining the water channel. Shading is used to distinguish chains in front and in the back. b, Schematic of the approximately hexagonal packing of several inverted-micelle cylinders. c, Cross sections through the cylindrical water channels (white) and the Nafion crystallites (black) in the noncrystalline Nafion matrix (dark gray), as used in the simulation of the small-angle scattering curves in d. d, Small-angle scattering data (circles) of Rubatat et al. in a log(I) vs. log(q) plot for Nafion at 20 vol% of H2O, and our simulated curve from the model shown in c. The inset shows the ionomer peak in a linear plot of I(q). Simulated scattering curves from the water channels and the crystallites by themselves (in a structureless matrix) are shown dashed and dotted, respectively.
Straightness of water channels in Nafion. The SAXS data have provided detailed information on the lateral packing of the channels, but knowledge on the behavior of the channels in the third dimension is limited. Specifically, the persistence length of the water channels, i.e. the length scale on which they are essentially straight, has not been determined reliably. An analysis of the narrowing of the 2H NMR spectrum of D2O in Nafion can contribute here. After initial narrowing due to exchange between free D2O in the interior of the channels and the D2O at the channel wall, the 2H quadrupolar coupling is averaged down to ~ 1 kHz for 10 wt% D2O, as observed in drawn Nafion samples (provided by Dr. Robert B. Moore, Virginia Tech) with almost straight channels. The coupling shows quantitatively the expected decrease with increasing dilution of bound water by free water.
If D2O diffuses along a curved channel on the millisecond time-scale, it experiences varying 2H quadrupolar couplings due the orientation dependence of the residual coupling, resulting in further motional narrowing. Such an additional reduction in line width by a factor > 10 is indeed observed in undrawn commercial Nafion membranes, see Figure 3(a) indicating that the channels are relatively tortuous on the micrometer scale probed by the diffusion with D ~ 1 µm2/ms. Treating a water channel as a chain of many short straight segments with different orientations, the diffusion of D2O can be numerically simulated as a multi-site exchange process. The preliminary simulations in Figure 3(b) show the expected dramatic line narrowing as the channels become more tortuous. The most stringent upper limit on the persistence length is provided by the 2H T2 relaxation time, which is also calculated in the simulations. For simulations of channels with short persistence lengths, a large number of segments must be used; the simulation time can be kept manageable by lumping several original segments together into an average-coupling segment.