5-fluorouracil has been a cornerstone of cancer treatment for decades, particularly against colorectal, breast, and gastrointestinal tumors. The drug works by inhibiting thymidylate synthase, an enzyme essential for DNA replication, thereby triggering apoptosis in rapidly dividing cancer cells. But the same biological potency that makes it effective in the clinic makes it hazardous once it escapes into the environment. Hospital effluents and surface waters are now routinely found to contain measurable concentrations of the compound, and conventional wastewater treatment systems are often unable to eliminate it completely. Because 5-fluorouracil is chemically stable and biologically active even at low concentrations, it poses risks to aquatic life and may contribute to the spread of drug-resistant microorganisms, creating an urgent need for removal strategies that actually work.

The materials at the heart of the new study are carbon coordination frameworks, conical carbon architectures with extended pi-conjugation that can be functionalized with remarkable precision. At the apex of each cone, the researchers embedded a porphyrin-like ring containing four nitrogen atoms and twenty carbon atoms, creating a cavity perfectly suited to host a single metal ion. This design prevents metal aggregation and ensures atomic dispersion, meaning every metal center remains individually accessible to incoming molecules. The team selected two representative metals for comparison: iron in its +2 oxidation state, a transition metal cation, and magnesium in its +2 state, a group IIA cation. Both were positioned at the central cavity of the nitrogen site to form the M-N4 carbon coordination frameworks that served as the computational testbeds.

Using the M06-2X exchange-correlation functional with the 6-31G(d) basis set in the Gaussian 16 package, the researchers systematically explored how 5-fluorouracil orients itself on these frameworks. Two stable adsorption configurations emerged for each metal, distinguished by which of the drug’s oxygen atoms coordinates to the metal center. The results were striking. The iron-based complexes bound the drug with adsorption energies of -30.89 and -28.90 kilocalories per mole for the two configurations, while the magnesium analogues managed only -25.23 and -22.19 kilocalories per mole. All values fell within the chemisorption regime, indicating the formation of moderately strong chemical bonds rather than weak physical attraction. The contact distances told the same story: the iron-oxygen separations measured approximately 1.99 angstroms, short enough to signal strong metal-ligand bonding, whereas the corresponding magnesium-oxygen distances were noticeably longer.

To understand why iron grips the drug so much more effectively, the team deployed a battery of electronic structure analyses. Atoms-in-Molecules theory examines the electron density at so-called bond critical points, the locations where electron density accumulates between interacting atoms. For the iron-oxygen contacts, the analysis revealed low electron density but a positive Laplacian paired with negative total energy density, a signature of partial covalency. The magnesium-oxygen contacts, by contrast, showed positive energy densities consistent with weak van der Waals interactions. Hydrogen bonds between a hydrogen atom of the drug and a nitrogen site of the framework provided additional stabilization in all complexes, with the strongest contact in the iron-based system reaching an estimated energy of -35.49 kilojoules per mole. Reduced density gradient analysis visualized these interactions directly, showing green iso-surfaces for the dispersive metal-oxygen contacts and blue regions marking the stabilizing hydrogen bonds.

Natural Bond Orbital analysis added another layer of mechanistic insight by quantifying the donor-acceptor interactions that stabilize the adsorbed complex. The dominant contribution came from charge donation from the oxygen lone-pair orbitals of the drug into antibonding orbitals of the iron center, while the interaction involving nitrogen lone pairs exhibited the largest second-order perturbation energy, indicating pronounced electron delocalization toward the metal. The charge-transfer parameter derived from conceptual density functional theory reached 0.0217 for the iron complex versus 0.0175 for the magnesium one, confirming that the iron framework accepts more electron density from the drug. Electrostatic potential maps made this exchange visible: the electron-rich oxygen atoms of the drug carry negative potential while the electron-deficient iron center is positive, and after adsorption the framework fragment gains negative character, direct evidence of charge flowing from drug to material.

Interestingly, the study also ruled out some potential applications. Adsorption of 5-fluorouracil induced only minor changes in the band gap of the frameworks, which measured 3.204 electron volts for the pristine iron system, confirming its semiconducting nature. The work function and the simulated ultraviolet-visible spectra likewise showed only slight variations upon drug binding. These findings imply that the frameworks would not function effectively as conductivity-based or optical sensors for detecting the drug, even though they bind it strongly. The metal-nitrogen bond lengths, however, did elongate measurably upon adsorption, suggesting that structural markers rather than electronic ones might serve as sensitive indicators of capture events in future experimental designs.

The research also looks toward the era of artificial intelligence. The authors benchmarked several graph neural network architectures, including GCN, GAT, MPNN, and GIN, for predicting band gaps within the QM9 chemical space, finding that the GIN architecture achieved the lowest prediction error at 0.082 electron volts mean absolute error. They emphasize that no machine-learning model was trained on their adsorption data, and the discussion is presented as a forward-looking perspective rather than a validated predictive framework. Still, the DFT-derived descriptors of adsorption energy, charge transfer, and electronic structure constitute a structured reference dataset that could seed future high-throughput screening campaigns, potentially accelerating the rational design of nano adsorbents without requiring exhaustive quantum calculations for every candidate material.

The practical takeaway is clear: iron-centered nitrogen-doped carbon coordination frameworks are markedly more promising than magnesium analogues for capturing 5-fluorouracil from contaminated water, thanks to shorter metal-oxygen contacts, partially covalent bonding, and superior charge-transfer capability. While the calculations were performed in the gas phase, supplementary polarizable continuum model calculations in water confirmed that the two adsorption configurations remain nearly isoenergetic and that the overall mechanism persists under aqueous conditions. As pharmaceutical pollution intensifies worldwide, studies like this one demonstrate how computational chemistry can shortlist the most effective materials before a single laboratory experiment is run, turning the abstract mathematics of electron density into a concrete blueprint for cleaner water.

Subject of Research: Density functional theory investigation of 5-fluorouracil adsorption on Fe(II)- and Mg(II)-centered N4 carbon coordination frameworks for pharmaceutical pollutant removal

Article Title: DFT investigation of 5-fluorouracil adsorption on Fe/Mg–N₄ carbon coordination frameworks: mechanistic insights

Article References: DFT investigation of 5-fluorouracil adsorption on Fe/Mg–N₄ carbon coordination frameworks: mechanistic insights. (n.d.). https://doi.org/10.1007/s44442-026-00086-6

Image Credits: AI Generated

DOI: 10.1007/s44442-026-00086-6

Keywords: 5-fluorouracil, DFT, carbon coordination frameworks, Fe-N4, Mg-N4, adsorption, chemisorption, AIM analysis, NBO analysis, water purification, pharmaceutical pollution, graph neural networks