Predicting Internal Versus External Nanoparticle Formation in Zr-Based Metal-Organic Frameworks

Metal nanoparticles supported on metal-organic frameworks (MOFs) can form either within the internal pore network or on the external crystal surface, creating distinct catalytic environments. Although both configurations are widely reported, nanoparticle spatial localization is typically treated as an empirical outcome rather than a predictable material property. Here, we investigate nanoparticle localization in Zr-based MOFs using a controlled double-solvent method to introduce metal precursors into the pore system. An experimental data set spanning 10 transition metals across 11 Zr-based MOFs was constructed, and nanoparticle positions were classified by transmission electron microscopy based on the presence or absence of detectable particles on external crystal surfaces. Combining metal descriptors describing framework affinity and surface mobility with a low-dimensional representation of MOF host chemistry, we identify a compact descriptor space that separates internal confinement from external nanoparticle formation. The resulting model reveals a localization boundary governed by the balance between metal affinity toward oxygen-containing nodes, metal mobility on π-conjugated linker environments, and linker heteroatom chemistry. External validation using previously unseen MOFs confirms the predictive capability of the model. These results establish nanoparticle spatial localization in MOF-supported catalysts as a predictable outcome determined by metal-framework interactions, guiding the control of nanoparticle position in porous catalytic materials.