(4ca) Multiscale Elucidation of Structure-Properties Relations for Molecular Transport in Polycrystalline Thin Films

Despite advances in computational power and
algorithm efficiency, invaluable fundamental insight from current molecular
models remains limited to a narrow range of short length and time scales.  In light of such limitations, a recent U.S.
Department of Energy (DOE) initiative [1] has put forth a roadmap directing development of
multiscale approaches to address science and engineering problems that are
currently ?out of reach? due to strongly coupled disparity in scales.  Among the roadmap targets is the rational
design of materials for applications such as molecule sensing, high-resolution
gas separations, and nanoscale materials templating.  Realization of such materials applications demands comprehensive
development of structure-properties relations. 
Concomitant challenges derive from the strongly coupled scales
associated with molecular transport in the presence of material
polycrystallinity (i.e., grain boundaries) and strong adsorbate-adsorbate forces. 

The ordered crystalline microstructure,
nanometer-sized pores, cation tunability, and established techniques for thin
film fabrication (e.g., [2]) make microporous materials (e.g., zeolites)
attractive for such applications.  The
current paradigm for modeling molecular transport in zeolite thin films focuses
almost exclusively on single-crystal systems characterized by relatively
weak adsorbate-adsorbate forces.  
Many systems of practical interest, however, are characterized by strongly
interacting adsorbates and substantial polycrystallinity.  This poster will provide an overview of
parallel theoretical and experimental efforts taken towards the development of
structure-properties relations for transport in realistic microporous thin
films:

I.  Multiscale modeling of diffusion of
strongly interacting molecular species in microporous membranes

Strong coupling between scales associated with
transport in polycrystalline microporous films plagues conventional molecular
simulations and leads to the break down of phenomenological continuum theories
(e.g., generalized Maxwell-Stefan).  In
this research, we develop and employ multiscale stochastic kinetic Monte Carlo
(KMC) methods [3] and continuum mesoscopic theories, e.g., [4-6], to bridge these disparities in scale.  The rich test-bed system of benzene in NaX
zeolite will be employed as an example application of these multiscale
techniques.  The approach first employs
a hierarchical technique to rationally parameterize complex molecular models of
host-guest systems [7].  Despite
limitations to thin (sub-micron) systems, subsequent gradient KMC simulations
of permeation through single crystal and polycrystalline membranes are in good
agreement with laboratory experiments (i.e., apparent activation energies of
permeation).  In addition, they reveal
molecular level insight into the role of crystal terminations, strong
adsorbate-adsorbate potentials [8], underlying diffusion mechanisms, and nanoscopic
defects in permeation through thin polycrystalline films.

To overcome the computational limitations of
conventional KMC, we derive a topology-specific mesoscopic model via rigorous
coarse-graining of the parameterized lattice representations.  The resulting continuum model retains
molecular level details of diffusion dynamics and adsorbate-adsorbate
interactions while accessing larger length and time scales required for
prediction of macroscopic permeation properties.  The accuracy of mesoscopic theories has been shown in the past
for a range of prototype systems (e.g., [4-6,
9]). Here, we assess the accuracy of these new
mesoscopic theories, derived for more complex systems, by direct comparison
with gradient KMC simulations in the limit of thin membranes and with
experimental permeation data for realistically thick membranes.  Ultimately, this multiscale approach yields
device-level mesoscopic models that begin to elucidate structure-properties
relations for these complex systems.

II.  Novel
non-destructive characterization of thin film polycrystallinity for elucidating
structure-properties relations

Our gradient KMC simulations predict substantial
sensitivity of permeation to only moderate membrane polycrystallinity.  This underscores the need for quantitative
characterization of microporous membrane polycrystallinity for development of
predictive models of membrane permeation. 
Fluorescence confocal optical microscopy (FCOM) studies [10, 11] involving selective adsorption of dye molecules in
polycrystalline features have highlighted the extent of this polycrystallinity
in zeolite membranes.  Quantitative
interpretation of FCOM images, however, has remained relatively elusive. 

This poster summarizes newly developed protocols for
pushing the limits of FCOM as a quantitative, non-destructive materials
characterization technique.  These
protocols involve rational screening of dye molecules via molecular mechanics
calculations for a priori assessment of steric compatibility and
energetic interactions with zeolitic (e.g., silicalite-1 and NaX) pores and
dominant crystal surfaces.  Novel
standards with nanometer scale features are also fabricated and imaged to
calibrate feature size and density with fluorescence intensity and its spatial
decay.  Such protocols, in conjunction
with correlative FCOM and SEM studies [11] using separate and sequential adsorption of dyes,
quantify the size and distribution of polycrystalline features, and link them
directly to the crystal morphology.  In
this poster, these protocols are applied to both siliceous silicalite-1 and, for the first time, NaX zeolite
membranes.  Comparison of characteristic
polycrystallinity is employed to explain variations in separation performance
associated with membrane orientation (in the case of silicalite-1) and growth
conditions.  Ultimately, we illustrate
the capabilities of more quantitative, non-destructive FCOM imaging for
elucidating structure-properties relations for polycrystalline zeolite
membranes.

Membranes for characterization were provided by
Prof. M. Tsapatsis and Dr. Z. Lai (Univ. of Minnesota) and Dr. V. Nikolakis
(Foundation for Research and Technology Hellas, Institute of Chemical
Engineering and High Temperature Chemical Processes, Patras, Greece).

References

1.         Dolbow, J., et al., Multiscale
Mathematics Initiative: A Roadmap. 2004, U.S. Department of Energy:
Washington, D.C.

2.         Lai, Z., et al., Science,
300 (2003), 456-460.

3.         Snyder, M.A., et al., Comput.
Chem. Eng., 29 (2005), 701-712.

4.         Vlachos, D.G. and M.A. Katsoulakis, Physical Review Letters, 85 (2000), 3898-3901.

5.         Snyder, M.A., et al., Chem.
Eng. Sci., 58(2003), 895-901.

6.         Lam, R., et al., Journal
of Chemical Physics, 115 (2001), 11278-11288.

7.         Snyder, M.A. and D.G. Vlachos, Molecular Simulation, 30 (2004), 561-577.

8.         Snyder, M.A. and D.G. Vlachos, Physical Review E, in press (2005).

9.         Snyder, M.A. and D.G. Vlachos, J. Chem. Phys., in press (2005).

10.        Bonilla, G., et al., J.
Membrane Sci., 182 (2001), 103-109.

11.        Snyder, M.A., et al., Microporous
and Mesoporous Materials, 76 (2004), 29-33.