Reports: ND1053140-ND10: Nanofabrication of Porous Membranes for Separations of Gaseous Energy Carriers Under Conditions of Single-File Diffusion

Kirk J. Ziegler, University of Florida

Sergey Vasenkov, University of Florida

Effective separation of light gases is important because of an increasing use of methane as a primary energy carrier. This work focuses on the development of a new separation strategy based on the single-file diffusion of select components in the gas mixture inside a porous membrane. Under these conditions, the effective diffusivity of a molecule is expected to decrease by many orders of magnitude in comparison to normal diffusion thus allowing for highly selective separations. The aim of this study is to fabricate model membranes capable of inducing single-file diffusion of molecules and on investigating diffusion in these membranes using pulsed field gradient (PFG) NMR at high field and high gradients. Membranes for separations under the conditions of single-file diffusion require the presence of parallel cylindrical channels with diameters in the range of one nanometer and relatively smooth walls. Our work during this year has focused on two approaches to achieve the desired dimensions:


(1) Fabricate porous membranes with dimensions below 10 nm. Highly ordered porous membranes of anodic aluminum oxide (AAO) were fabricated using 2-step anodization in selenic acid. As shown in Figure 1, the mean pore diameter was measured to be 9.4 nm, which is a noticeably smaller distribution in size than previously reported values (range of 7.1 – 13.4 nm). During the upcoming year, further reductions in pore diameter will be achieved by the synthesis of carbon nanotubes within the pore channels. The pore diameter will be controlled by the number of shells of the CNT. The shell walls will also be atomically flat, which is important for observing single-file diffusion.

(2) Fabricate sacrificial nanowires for pore formation via monodisperse catalyst particle arrays. Although current approaches achieve monodisperse catalyst particles on a surface, the particles tend to coalesce and undergo Ostwald ripening during the growth of nanowires. Therefore, the dimensions of the nanowires rarely match the initial diameter of the particles. To overcome these problems, highly ordered AAO membranes are used to alter the substrate surface to yield uniformly sized and spaced catalyst particles within nanocavities on the surface. As illustrated in Figure 2a, the AAO membrane (shown in Figure 2b and 2c) was first synthesized. A layer of polystyrene is then coated as a protective layer onto the AAO membrane before transfer to a Si substrate. Reactive ion etching of the substrate using the AAO membrane as the etch mask, followed by dissolution of the AAO membrane yields nanocavities on the Si surface. As shown in Figure 2d and 2f, the closely packed hexagonal pattern was transferred onto the underlying substrate with identical density and pore size.


Nanoparticles can be easily deposited into the nanocavities by either (i) deposition and annealing of a film or (ii) by spin-coating a nanoparticle suspension. For example, a thin layer of Au was sputtered into the nanocavities shown in Figure 3a. As shown in Figure 3b, the deposition and annealing of the Au after AAO removal yields Au nanoparticles sitting at the bottom of nanocavities. The size and spacing of these nanoparticles was dependent on the pore size and structure of the AAO membrane as well as the thickness of the sputtered Au film. Alternatively, a suspension of iron oxide nanoparticles with uniform diameters was spread onto the nanocavities. As shown in Figure 4, the iron oxide catalyst particles residing in each nanocavity has nearly the same diameter and most of the nanocavities contain a single nanoparticle. The new process developed to fabricate nanocavities in Si substrates may allow for exceptional control over the diameter of the catalyst particles and the nanowires grown from them. A membrane for single-file diffusion can then be fabricated using these sacrificial nanowires. Further reductions in pore diameter can be achieved by oxidation and etching of the nanowires.

Preliminary measurements of single-file diffusion in nanoporous membranes. PFG-NMR at high field and high gradients is uniquely suited for distinguishing between single-file and normal diffusion. These studies have focused on model systems prior to studies on the membranes fabricated above. C-13 and Xe-129 PFG-NMR at high magnetic field of 17.6 T and large magnetic field gradients up to 30 T/m was applied to study the diffusion of CO in L-Alanyl-L-Valine (AV) dipeptide nanotubes and Xe in phenylethynylene bis urea nanotubes. The most interesting results were obtained for diffusion of CO molecules in AV nanotubes at 298 K for a broad range of diffusion times between 30 and 500 ms. The CO loadings in the nanotubes corresponded to sorption equilibrium with 3 and 10 bar of CO in the gas phase. Our preliminary PFG-NMR data indicate that the mean square displacements (MSD) of CO molecules at different diffusion times exhibit behavior that is intermediate between single-file and normal diffusion. Single-file diffusion is expected in narrow one-dimensional channels where sorbate molecules cannot pass one another. For sufficiently long channels, this leads to increased MSD as a function of the square root of the diffusion time. In contrast, MSD is proportional to diffusion time for normal diffusion because sorbate molecules can pass one another. Under our measurement conditions, the nanotube radius (around 0.25 nm) was comparable to the smallest dimension of CO molecules. Hence, it is possible that mutual passages of CO molecules are not completely excluded in AV nanotubes. This can explain the observed time dependence of MSD, which appears to be intermediate between those corresponding to the single-file and normal diffusion. PFG-NMR studies of mixtures of CO and larger sorbate molecules, such as methane and Xe, in AVs will be carried out in the near future. In these mixtures, the smaller sorbate (CO) is expected to change to single-file diffusion because the presence of the larger sorbate species (methane or Xe) would prevent mutual passages of CO and the larger sorbate in the nanotubes. If successful, these studies will demonstrate induction of single-file diffusion of sorbate molecules. Such single-file induction can be used in highly-selective separations of gas mixtures.