Pore network set as input: inverse design of amorphous nanoporous carbon matches reference within 5 %
In a simulation study, pores of a target size distribution are set first and carbon is grown around them; density, surface area and pore volume match a quenched-MD reference within about 5 %.
Authors of an arXiv · preprint, not yet peer reviewed, describe a way to build models of disordered nanoporous carbon in which the pore network is specified first. Pores drawn from a target size distribution fill a prescribed volume fraction, carbon is grown around them by stochastic ring placement, and a particle swarm search recovers the construction parameters from a structural fingerprint. Against a carbon produced by quenched molecular dynamics (QMD), the search reproduces density, surface area and pore volume within about 5 %, together with the pore size distribution arxiv.org
arXiv · препринтarxiv.orgOpen source ↗.
Why pore geometry matters
Porous carbons are used in supercapacitor electrodes, gas storage and separation media, and water purification, and their performance depends on the geometry of the pores. The authors cite work showing that capacitance is set by the pore size distribution and not by surface area alone. In one cited case, two carbons with nearly the same distribution differed by 43 % in capacitance, which the original authors attributed to local structure arxiv.org
arXiv · препринтarxiv.orgOpen source ↗. A useful model therefore needs both a specified void geometry and the local structure of a disordered carbon.
Existing approaches cover only part of this. Slit-pore models have exact geometry, but their skeleton lacks the curvature and non-hexagonal rings of disordered carbon. Reconstruction methods need measurements of a sample that already exists, and their inversion is non-unique. Mimetic molecular dynamics yields a physically relaxed structure, but the pore network is an output that depends on the density, the thermal protocol and the interatomic potential arxiv.org
arXiv · препринтarxiv.orgOpen source ↗.
What the search does
As a stress test, the method also recovers a bimodal pore size distribution. According to the authors, the search history shows which parameters the targets constrain and flags a failed search arxiv.org
arXiv · препринтarxiv.orgOpen source ↗.
The constructed network is then completed by a short anneal. It recovers the coordination of the reference carbon, while the pore size distribution keeps its shape arxiv.org
arXiv · препринтarxiv.orgOpen source ↗.
Molecular dynamics still supplies the local chemistry in this scheme. The difference is that the pore geometry becomes a design variable chosen for a given application, rather than a result of the synthesis simulation arxiv.org
arXiv · препринтarxiv.orgOpen source ↗. The results come from a preprint and have not been peer reviewed.
