David Gryc

Framework topology and particle size jointly govern pore architecture, diffusion limitations, and accessible surface area, making their simultaneous control essential for functional metal–organic framework (MOF) design. Achieving reproducible control over both remains a central challenge in MOF chemistry, particularly for systems capable of crystallizing in multiple polymorphs. Here, we systematically explore the synthesis of two Zr‐based MOFs, NU‐1000 and NU‐901, which share the same Zr ₆ ‐oxo cluster and pyrene‐tetrabenzoate linker but differ in topology. To remove variability arising from in situ precursor formation, we employed preformed Zr ₆ ‐oxo clusters as well‐defined nodes. Building on this controlled starting point, multiple particle‐size regimes (<100 nm to ∼1 µm) were accessed through systematic variation of modulator identity and loading, co‐modulator content, linker equivalents, temperature, and dilution. Machine‐learning (ML) analysis quantitatively disentangles the multivariate influences governing phase selection and particle‐size evolution. This integrated approach enables predictive navigation of the synthesis parameter space, providing reproducible access to MOFs with defined size regimes and controlled topology. Together, these results establish a generalizable experimental‐computational workflow that transforms empirical MOF synthesis into a predictable, designable process, enabling rational control over polymorphism and growth in structurally adaptable framework materials.

Metal–organic frameworks (MOFs) are versatile precursors for nanostructured, heteroatom-doped carbons. However, understanding the spatiotemporally resolved mechanisms that govern crystalline-to-amorphous framework collapse, diffusion-mediated reorganization of transient intermediates, and the subsequent formation of graphitic carbon remains challenging. Here, we elucidate the transformation of the 2D leaf-shaped zeolitic imidazolate framework ZIF-L into Zn-dispersed N-doped carbon using a combination of in situ transmission electron microscopy (TEM) and X-ray diffraction (XRD). These measurements directly capture the sequential collapse of the layered MOF framework and the associated morphological evolution under controlled heating, decoupled from beam-induced effects during electron microscopy. Complementary ex-situ pyrolysis (700–1100 °C), together with high-resolution TEM, synchrotron X-ray pair distribution function (X-PDF) analysis, and electron energy-loss spectroscopy (EELS), reveals the diffusion of Zn-containing species, ligand decomposition, and progressive graphitization within the amorphous matrix. We identify temperature-dependent stages involving ligand breakdown, migration of Zn–Nx species, followed by Zn loss, and ultimately forming porous N-doped carbon at 1000 °C that exhibits the strongest electrochemical response within this series. These findings establish temperature-dependent structure–property relationships that enable the thermal engineering of MOF-derived ORR catalysts.

Despite extensive investigation of metal-organic framework (MOF) derived materials over the last 20 years, no systematic approach to predict the structural properties of the derived metal oxides is available. We present an integrated machine learning (ML) approach leveraging Smooth Overlap of Atomic Positions (SOAP) and multiple ML models, including Kernel Ridge Regression (KRR), to predict thermally derived zirconium dioxide (ZrO2) polymorph from diverse Zr-based precursors. By a systematic experimental dataset of calcination parameters, we train the ML model to quantitatively forecast material properties and the weight fraction of crystalline phases of the resulting ZrO2. Experimental validation of model predictions confirms that the chemical composition of precursors and calcination parameters have a profound influence on the crystallinity of MOF-derived ZrO2 polymorph. Our findings demonstrate the utility of small-data-driven predictive ML modeling and transfer learning for guiding the synthesis of advanced oxide materials providing a blueprint for accelerated discovery of MOF-derived nanomaterials.

Metal‐organic framework (MOF)‐derived materials are promising candidates for energy storage applications, particularly as supercapacitors, owing to their high porosity, tunable compositions, and pseudocapacitive behavior. However, performance optimization traditionally relies on resource‐intensive trial‐and‐error methods with limited insight into complex structural transitions. To address this challenge, a machine learning approach integrating Bayesian optimization (BO) to refine synthesis parameters systematically is presented. The importance of each parameter is assessed using correlation matrices, surrogate model structure, and Shapley values analysis in two models: random forest regressor, which achieves low prediction error, and extra trees regressor, which provides better generalization. This strategy efficiently explores the design space and significantly reduces experimental workload. Focusing on Mn‐MIL‐100‐derived MnO/C composites for supercapacitor applications, this approach identifies optimal conditions to increase energy storage performance while quantifying the influence of key process parameters. These findings demonstrate the potential of AI‐driven strategies to accelerate material discovery, enhance process efficiency, and advance the practical applications of MOF‐derived materials.