PFAS are the fluorinated molecules that don't break down: in rainwater, in blood, in fish. One way to pull them out of water is a metal-organic framework — a crystal of metal clusters and organic struts with pores a molecule can be caught in. This page builds one such cage in your browser, drops a PFAS ion inside and searches for where it settles: with a simple force field, in ångströms, and nothing synthesised.
Candidate
—
not screened
—
N⁺···O(PFO⁻)
—
not screened
—
N⁺···O(PFO⁻)
Zr₆ nodes and linkersN⁺ sitefluorinePFO⁻ guest
Pyrene-2,7-dicarboxylate: a flat, four-ring strut, the same length as biphenyl. In every candidate, half of the twelve linkers carry a trimethylammonium site (N⁺) to hold the ion's carboxylate head. That is the “50”.
A UiO-67-type Zr₆ framework: one octahedral cage, six nodes 19.0 Å apart along twelve linkers, every heavy atom placed from textbook bond lengths — C330 N6 O192 Zr36. Hydrogens and counter-ions are left out. Drag to turn it, scroll to zoom.
Dry-lab screening
For each candidate the PFO⁻ ion is dropped into the cage in 600 random positions and orientations, and the 6 best are refined step by step. The score is a simple force field: Lennard-Jones contacts with UFF parameters, plus point charges on the N⁺ sites and the carboxylate with a dielectric that grows with distance. Guest and cage are rigid. Same seed, same answer.
Results
Cage
kcal/mol
eV
N⁺···O Å
closest Å
< 4 Å
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
Lower is stronger. These numbers are this model's: not a measurement, and not the DFT figures in the reference. They rank three cages under one force field, and that is all they do.
What is PFAS?
Per- and polyfluoroalkyl substances: carbon chains dressed in fluorine. The carbon–fluorine bond is about the strongest in organic chemistry, so these molecules survive water treatment, sunlight and digestion, and they have been made since the 1940s for non-stick pans, firefighting foam and waterproof coats. The one in this cage is perfluorooctanoate, PFO⁻: an eight-carbon chain with fifteen fluorines and a carboxylate head, which carries a negative charge in water.
That charge is the handle. Something positively charged, held in a pore of the right size, can grab the head while the fluorinated tail packs against the pore's walls.
Capturing it with a MOF
A metal-organic framework is a crystal of metal clusters joined by organic linkers, with more empty space than solid: a gram of UiO-67 has the surface area of a football pitch inside it. Here the clusters are Zr₆O₄(OH)₄ octahedra, each binding twelve carboxylates, and the linkers are the struts between them. Swap the linker and the pore changes; decorate it and the pore's chemistry changes.
The three candidates try exactly that. All carry trimethylammonium sites on half their linkers, to hold the carboxylate head; they differ in the strut — a flat pyrene, a biphenyl with fluorinated groups on the walls, and plain biphenyl as the control. The screening asks which cage the ion settles into most deeply.
What runs here, and what doesn't
Geometry from bond lengths
The cage is assembled from textbook distances: Zr–O 2.23 Å, 1.39 Å for an aromatic C–C, a biphenyl twist of 30°. They close on UiO-67's measured cell, 26.9 Å, which is the check that the picture is a structure. Nothing has been relaxed, hydrogens are omitted, and the counter-ions that would balance the N⁺ sites are absent.
A simple force field
Lennard-Jones contacts with UFF parameters, point charges on the N⁺ sites and on the guest, a dielectric that grows with distance, a rigid guest and a rigid cage. It is the kind of score a first pass in a real dry lab uses to throw out the hopeless. It is not density-functional theory, and nothing here should be quoted as an adsorption energy.
Not the reference, and not a lab
This page was prompted by Xiaomi's write-up on using MiMo-V2.6-Pro in materials R&D, which screened UiO-67-type cages for PFO⁻ with DFT. Their candidates, structures and energies are theirs and are not reproduced here; these cages were built from scratch for this page. Whether any of the three would work is a wet-lab question.