|
Approximately 75% of electricity used
in the United States
is produced by coal-burning power plants that spew carbon dioxide into the
atmosphere and contribute to global warming. To reduce this effect, many
researchers are searching for porous materials to filter out the carbon dioxide
generated by these plants before it reaches the atmosphere, a process commonly
known as carbon capture. But identifying these materials is easier said than
done.
“There are a number of porous
substances—including crystalline porous materials, such as zeolites, and
metal-organic frameworks—that could be used to capture carbon dioxide from
power plant emissions,” says Maciej Haranczyk, a scientist in the Lawrence
Berkeley National Laboratory’s (Berkeley Lab) Computational Research Division.
In the category of zeolites alone,
Haranczyk notes that there are around 200 known materials and 2.5 million
structures predicted by computational methods. That’s why Haranczyk and
colleagues have developed a computational tool that can help researchers sort
through vast databases of porous materials to identify promising carbon capture
candidates—and at record speeds. They call it Zeo++.
By using Zeo++ researchers have already
sifted through one such database of millions of materials and have identified a
few that could outperform current technologies.
The
tool, he notes, works not by simulating each atom of a material, but by mapping
what isn’t there: The voids in the materials.
Taking the
molecule’s eye view
Porous materials like zeolites or metal-organic frameworks come in a variety of
shapes and have a range of pore sizes. It is actually the shape and pore sizes
that determine which molecules get absorbed into the material and which ones
pass through.
Like molecular sponges, porous
materials can also be reused in a cycle of capture and release. For instance,
in the case of carbon capture, once the material is saturated and cannot absorb
any more carbon dioxide, the gas can be extracted, and the cycle repeated.
“Understanding how all of these factors
combine to effectively capture carbon is a challenge,” says Richard Luis
Martin, a member of the Zeo++ development team and a postdoctoral researcher in
Berkeley Laboratory’s Computational Research Division. “Until Zeo++, there were
no easy methods for analyzing such large numbers of material structures and
identifying what makes a material an outstanding carbon catcher.”
He notes that silicious zeolites, to
take one example, are composed of the same tetrahedral blocks of silicon and
oxygen atoms, but the geometric arrangement of these blocks differs from one
zeolite to the next, and this configuration is what determines how carbon
dioxide or any other molecule will interact with the porous material.
Before Zeo++, scientists would
typically characterize a porous structure based on a single feature, like the
size of its largest pore or its total volume of empty space, then compare and
categorize it based on this single observation.
“The problem with this 1D description
is that it does not tell you anything about how a molecule like carbon dioxide
will move through the material,” says Martin. “To identify the most effective
materials for absorbing carbon dioxide, we need to understand the porous
structure from the perspective of the penetrating molecule.”
This is precisely why Zeo++
characterizes these structures by mapping the empty spaces between their atoms.
Drawing from a database of the coordinates of all the atoms in each porous
structure, Zeo++ generates a 3D map of the voids in each material. This 3D
network allows researchers to see where the channels between atoms intersect to
create cavities. The size and shape of these cavities determine whether a
molecule will pass through the system or be absorbed.
Using a tool called Voronoi Holograms,
also developed by the Berkeley Laboratory team, researchers can automatically
compare these 3D maps to identify materials with similar pore sizes and
structures.
“Using this technique we can examine
the pathways between atoms and see how these paths connect to create a larger
network,” Martin says.
Since researchers already know the
positions of atoms in materials being considered, the tool can chart these
empty channels relatively quickly. And because researchers are not dealing with
all of the atoms in a structure, but a more simple representation of the
material’s empty space, Zeo++ can run its analysis much faster and with a lot
less computing power than a typical simulation based on physical models.
For
their first project, the team analyzed a database of millions of predicted
porous material structures compiled by Rice University Professor Michael Deem
and his research group to identify which would be most effective for trapping
molecules like carbon dioxide.
Narrowing the field faster
Searching vast sets of materials for desired characteristics is a problem that
is not unique to the quest for carbon capture. In fact, the pharmaceutical
industry faces similar challenges in the search for new drugs, and uses
informatics approaches to explore large databases of drug candidates. Haranczyk
and his Berkeley Laboratory colleagues applied some of these industry concepts
combined with state-of-the-art advances in mathematics and computational
algorithms to create Zeo++.
According to Berend Smit, who leads the
Energy Frontier Research Center
for Gas Separations Relevant to Clean Energy Technologies at the University of
California at Berkeley, it used to take weeks to manually characterize and
compare the porous structures of a mere 20 materials. With Zeo++, researchers
can analyze an entire database, containing information on hundreds-of-thousands
to millions of porous structures, in a few days.
“Zeo++
allows us to do things that would otherwise be physically impossible,” says
Smit, whose group is developing laboratory and computational methods for
identifying carbon dioxide-absorbing nanomaterials.
