SEARCH · Search NASA
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
The host–guest chemistry between a series of anionic MOFs and their trapped counterions was investigated by single crystal XRD.
Predicting creep crack growth (CCG) of flaws found during operation in high-temperature alloy components is essential for assessing the remaining lifetime of those components. While defect assessment procedures are available for this purpose in design codes, these are limited in their range of applicability. This study assesses the application of a local damage-based finite-element methodology as a more general technique for the prediction of CCG at high temperatures on a variety of structural configurations. Numerical results for stainless steel 316H, which are validated against experimental data, show the promise of this approach. This integration of continuum damage mechanics (CDM) based methodologies, together with adequate inelastic models, into assessment procedures can therefore inform the characterization of CCG under complex operating conditions, while avoiding excessive conservatism. This article shows that such modeling frameworks can be calibrated to experimental data and used to demonstrate that the degree of tri-axiality ahead of a growing creep crack affects its rate of growth. The framework is also successfully employed in characterizing CCG in a realistic reactor pressure vessel geometry under an arbitrary loading condition. These results are particularly relevant to the nuclear power industry for defect assessment and inspections as part of codified practices of structural components with flaws in high-temperature reactors.
Important mechanisms that lead to features, often complex, in X-ray photoelectron spectroscopy (XPS) spectra are defined and described. It is shown that there is much information in an XPS spectrum that can be obtained by examining these features rather than examining only the shifts of main peaks between different materials. These mechanisms are presented with a focus on describing the underlying chemical and physical phenomena responsible for features of the XPS and on showing how these XPS features can be related to the properties and electronic structure of the material studied. While it is necessary to consider certain quantum mechanical rules, the mathematical formalism is not discussed. However, a general awareness of multiplet splittings, which are a result of angular momentum coupling combined with ligand field and spin-orbit splittings, and of covalent mixings in the metal-ligand bond of oxides is essential to properly interpret the significance of XPS features. A conceptual framework of shake excitation from bonding to anti-bonding orbitals is introduced to provide an understanding of the significance of XPS satellites. While the coupling of theory and measurement is required to extract quantitative information from XPS, it may be possible to obtain useful qualitative information directly from features of the XPS spectra provided that one takes into account more than only shifts of the XPS binding energies.
An efficient numerical framework is presented for modeling viscoelasticity and permanent set of polymers. It is based on the hereditary integral form of transient network theory, in which polymer chains belong to distinct networks each with different natural equilibrium states. Chains continually detach from previously formed networks and reattach to new networks in a state of zero stress. The free energy of these networks is given in terms of the deformation gradient relative to the configuration at which the network was born . A decomposition of the kernel for various free energies allows for a recurrence relationship to be established, bypassing the need to integrate over all time history. The technique is established for both highly compressible and nearly incompressible materials through the use of neo-Hookean, Blatz–Ko, Yeoh, and Ogden-Hill material models. Multiple examples are presented showing the ability to handle rate-dependent response and residual strains under complex loading histories.
Inducing porosity in solid coordination complex crystals, which are an important class of catalysts, is critical for many applications where a high surface area is required. However, unlike metal organic frameworks (MOFs), fabrication of porous coordination crystals remains a significant challenge. Here we demonstrate a simple method to produce and modulate porosity in coordination complex crystals using ferroin perchlorate, a model system that combines a common ionic complex with a very reactive counter-ion. Using thermal chemical vapor deposition (CVD), we show that by annealing ferroin perchlorate crystals at 350 °C under a flow of ethylene, hydrogen, argon, and oxygen, we induced pores in the crystal. We demonstrate that small amounts of oxygen, which may combine with hydrogen to form water, are essential for pore formation. We also demonstrate that pore size and density can be easily controlled by varying the ethylene flow. Upon raising the annealing temperature to 500 °C, we observed a second transition in which carbon nanofibers (CNFs) grew from the porous crystal. This approach represents a simple and effective method for the synthesis of porous materials with good control over pore size and density. It also enables the synthesis of complex networks of nanostructures (in our case CNFs) by simply varying process parameters such as temperature and gas flows. This represents an important advance for the fabrication of porous coordination complex crystals. - Graphical abstract: Annealing ferroin perchlorate at 350 °C induces porosity in the complex while annealing at 500 °C leads to the additional growth of carbon nanofibers Display Omitted.
The development of homogeneous catalysis is enabled by the availability of a rich toolkit of kinetics experiments, such as the Hgdrop test, that differentiate catalytic activity at ligand-supported metal complexes from potential heterogeneous catalysts derived from decomposition of molecular species. Metal-organic frameworks (MOFs) have garnered significant attention as platforms for catalysis at site-isolated, interstitial catalyst sites. Unlike homogeneous catalysis, relatively few strategies have been advanced to evaluate the origin of catalytic activity in MOF-promoted reactions. Many of the MOFs that have been examined as potential catalysts are comprised of molecular constituents that represent viable catalysts in the absence of the extended MOF lattice, and thus interfacial sites and leached homogeneous species represent potential sources of catalyst activity. Here, we demonstrate that analysis of deuterium kinetic isotope effects (KIEs) and olefin epoxidation diastereoselectivity provides probes of the origin of catalytic activity in MOF-promoted oxidation reactions. These analyses support involvement of lattice-based Fe sites in the turnover-limiting step of C–H activation with Fe-MOF-74-based materials (i.e., the MOF functions as a bona fide catalyst) and evidence that Cu 2 -based MOF MIL-125-Cu 2 O 2 functions as a solid-state initiator for solution-phase oxidation chemistry and is not involved in the turnover limiting step (i.e., the MOF does not function as a catalyst for substrate functionalization). furthermore, we anticipate that the simple experiments described here will provide a valuable tool for clarifying the role of MOFs in C–H oxidation reactions.
A large number of Zintl phases have been discovered by solid-state chemists driven by empirical knowledge, chemical intuition and in some cases, through serendipitous accidents. These discoveries have only scratched the surface, given the vast compositional and structural diversity that Zintl phases can accommodate. The large chemical space of Zintl phases, as well as intermetallic compounds in general, remain under-explored. Here, we use graph neural networks and the upper bound energy minimization approach to efficiently scan a large chemical space of >90 000 hypothetical Zintl phases and accurately discover 1810 new thermodynamically stable phases with 90% precision, as validated with first-principles calculations. We show that our approach is more than 2× more accurate in predicting DFT stability than M3GNet (40% precision) on the same dataset. Using a random forest model and SHAP analysis, we demonstrate the critical role of ionic bonding in the thermodynamic stability of Zintl phases. Our results not only expand the known chemical landscape of Zintl phases but also highlight the efficacy of machine learning frameworks combined with domain knowledge in uncovering chemically meaningful insights across complex intermetallics.
In December of 2013, the U.S. Department of Energy’s SSL R&D Program released a Funding Opportunity Announcement (FOA), that for the first time, contained opportunities for comprehensive application-specific system development. The FOA included opportunities for two applications, one of which was a Patient Room. Philips Lighting Research North America, submitted a proposal for the Patient Room application, and was selected for the complete project award. The award amount was for $497,127, with a Philips Research co-funding commitment 165,709 dollars. The total project value was 662,836 dollars. This project sought to redefine lighting for the patient room application. The goal was to deliver an innovative LED patient suite (patient room and bathroom) lighting system solution that was 40% more energy-efficient than traditional fluorescent incumbent technologies, and would meet all the visual and non-visual needs of patients, caregivers and visitors, and improve the patient experience. State-of-the-art multichannel LED platforms and control technologies that would provide spectral tuning and become part of an intelligent, connected lighting system drove the solution. The project was scoped into four main task areas that included a) System Concept Creation, b) Identification of the Luminaire Portfolio, c) Development of the Connected Lighting Infrastructure, and d) System Performance Validation. Each of the four main tasks were completed and validated extensively over the course the 2 ½ year project. The system concept was created by first developing a lighting design that demonstrated best practices for patient room lighting – illuminance and uniformity for task performance, reduced glare, and convenient controls, in addition to giving patients control over the lighting in their environment. A framework was defined to deliver circadian support via software behaviors. Through that process luminaires were identified from the Philips portfolio that were adaptable – by their form, dimensions, and optical materials – to mix multicolor LED platforms uniformly and deliver target design lumen levels. The Blue Sky luminaire was selected for the patient bed area to give the illusion of skylight while providing white light on the patient bed. Luminaires used existing 2-channel tunable white LED boards, and newly developed 4-channel LED boards. Red-Orange, Blue, Green, and Blue-shifted Yellow LED chips were selected based on spectral characteristics and their ability to produce high quality white light. 4-channel Power over Ethernet (PoE) drivers were developed and firmware written so they would communicate with both 2- and 4-channel boards. These components formed the backbone of the connected lighting infrastructure. Software, flexible and nuanced in its complexity, was written to set behaviors for myriad lighting scenes in the room throughout the 24 hour day – and all could be overridden by manual controls. This included a dynamic tunable white program, three color changing automatic programs that simulated degrees of sunrise to sunset palettes, and an amber night lighting system that offered visual cues for postural stability to minimize the risk of falls. All programs were carefully designed to provide visual comfort for all occupants, support critical task performance for staff, and to support the patient’s 24hr rhythms. A full scale mockup room was constructed in the Philips Cambridge Lab. The lighting system was installed, tested and functionality demonstrated to ensure smooth operation of system components – luminaires, drivers, PoE switches, wall controls, patient remote, and daylight and occupancy sensors. How did the system perform? It met visual criteria, confirmed by calculations, simulations and measurements in the field. It met non-visual criteria, confirmed by setting circadian stimulus (CS) targets and performing calculations using the calculator developed by the Lighting Research Center. Finally, human factors validation studies were conducted to gain insight from real end users in the healthcare profession; surveys were administered, data analyzed, and audio comments captured. The general consensus was positive, with requests to pilot the system in their hospitals. The importance of the research completed under this grant is that it allowed the exploration and development of a unique lighting system, one that would deliver a blend of visual and non-visual criteria in patient room design for today’s healthcare environment. The research investigated the area of multichannel LED technology, multichannel Power over Ethernet (PoE) drivers and their integration with automatic and manual controls as a system – uncovering and meeting challenges along the way. It married visual needs of patients and staff with support for 24 hour rhythms, placing value on the wellbeing of the patient – while successfully saving energy over incumbent technologies. Indications are that the market is ready and willing to invest – multiple healthcare facilities are in line to pilot this system, recognizing its value beyond energy to patient and staff well-being. Its value to the public can best be expressed by a patient support coordinator who, after spending several hours in the room being immersed in the lighting, analyzing all its features, commented: “This re-writes lighting for healthcare”.
In this project we studied both natural photosynthetic antenna complexes and various artificial systems (e.g. chlorophyll (Chl) trefoils) using high resolution hole-burning (HB) spectroscopy and excitonic calculations. Results obtained provided more insight into the electronic (excitonic) structure, inhomogeneity, electron-phonon coupling strength, vibrational frequencies, and excitation energy (or electron) transfer (EET) processes in several antennas and reaction centers. For example, our recent work provided important constraints and parameters for more advanced excitonic calculations of CP43, CP47, and PSII core complexes. Improved theoretical description of HB spectra for various model systems offers new insight into the excitonic structure and composition of low-energy absorption traps in very several antenna protein complexes and reaction centers. We anticipate that better understanding of HB spectra obtained for various photosynthetic complexes and their simultaneous fits with other optical spectra (i.e. absorption, emission, and circular dichroism spectra) provides more insight into the underlying electronic structures of these important biological systems. Our recent progress provides a necessary framework for probing the electronic structure of these systems via Hole Burning Spectroscopy. For example, we have shown that the theoretical description of non-resonant holes is more restrictive (in terms of possible site energies) than those of absorption and emission spectra. We have demonstrated that simultaneous description of linear optical spectra along with HB spectra provides more realistic site energies. We have also developed new algorithms to describe both nonresonant and resonant hole-burn spectra using more advanced Redfield theory. Simultaneous description of various optical spectra for complex biological system, e.g. artificial antenna systems, FMO protein complexes, water soluble protein complexes, and various mutants of reaction centers continues; this work is supported by the new DOE BES grant.
In order to establish quantitative process–structure–property relationships in thermal spray coatings, a robust framework for defining (micro)structural characteristics is needed. Here, we present a quantitative characterization of the three-dimensional morphology of porosity in spray-formed tantalum samples based on high-resolution X-ray micro-computed tomography. Using synchrotron facilities, we acquired dozens of high-resolution scans, enabling a statistically meaningful comparison across multiple samples, different regions within samples, and spray processes. We quantify the spatial distribution, size, and topology of porous inclusions, with a significant focus on variability across samples and different spray processes (plasma and cold sprayed), as well as sensitivity to image segmentation and resolution. Based on a typical segmentation, we report porosities ranging from 0.9 to 1.7 pct for all samples tested, with significant sensitivity due to image segmentation resulting in estimates as low as 0.6 pct and as high as 4.8 pct. For the complex pore space morphology observed in these materials, we argue that a conventional analysis based on identifying individual pores is not well suited, and propose an alternative approach based on morphological metrics with a rich history in porous media literature, such as spatial correlations, local pore thickness, and scale-dependent sub-sampling. Spatial correlations indicate anisotropic splat structures, but only mildly anisotropic pores. Various measures of pore size show a wide distribution of sizes, ranging from sub-micron to 10-micron length scales. Scale-dependent variations in porosity suggest that representative volumes of several hundred microns are required for convergence of morphological metrics, with larger volumes for cold-spray materials. This work provides a robust quantitative basis for describing three-dimensional pore structure in thermal spray coatings.
An acute reduction in oxygen (O2) delivery to a tissue is generally associated with a decrease in phosphocreatine, increases in ADP, NADH/NAD, and inorganic phosphate, increased rates of glycolysis and lactate production, and reduced rates of pyruvate and fatty acid oxidation. However, given the complexity of the human bioenergetic system and its components, it is difficult to determine quantitatively how cellular metabolic processes interact to maintain ATP homeostasis during stress (e.g., hypoxia, ischemia, and exercise). Of special interest is the determination of mechanisms relating tissue oxygenation to observed metabolic responses at the tissue, organ, and whole body levels and the quantification of how changes in tissue O2 availability affect the pathways of ATP synthesis and the metabolites that control these pathways. In this study, we extend a previously developed mathematical model of human bioenergetics to provide a physicochemical framework that permits quantitative understanding of O2 as a metabolic regulator. Specifically, the enhancement permits studying the effects of variations in tissue oxygenation and in parameters controlling the rate of cellular respiration on glycolysis, lactate production, and pyruvate oxidation. The whole body is described as a bioenergetic system consisting of metabolically distinct tissue/organ subsystems that exchange materials with the blood. In order to study the dynamic response of each subsystem to stimuli, we solve the ordinary differential equations describing the temporal evolution of metabolite levels, given the initial concentrations. The solver used in the present study is the packaged code LSODE, as implemented in the NASA Lewis kinetics and sensitivity analysis code, LSENS. A major advantage of LSENS is the efficient procedures supporting systematic sensitivity analysis, which provides the basic methods for studying parameter sensitivities (i.e., changes in model behavior due to parameter variation). Sensitivity analysis establishes relationships between model predictions and problem parameters (i.e., initial concentrations, rate coefficients, etc). It helps determine the effects of uncertainties or changes in these input parameters on the predictions, which ultimately are compared with experimental observations in order to validate the model. Sensitivity analysis can identify parameters that must be determined accurately because of their large effect on the model predictions and parameters that need not be known with great precision because they have little or no effect on the solution. This capability may prove to be important in optimizing the design of experiments, thereby reducing the use of animals. This approach can be applied to study the metabolic effects of reduced oxygen delivery to cardiac muscle due to local myocardial ischemia and the effects of acute hypoxia on brain metabolism. Other important applications of sensitivity analysis include identification of quantitatively relevant pathways and biochemical species within an overall mechanism, when examining the effects of a genetic anomaly or pathological state on energetic system components and whole system behavior.
Inverse problems constrained by partial differential equations (PDEs) play a critical role in model development and calibration. In many applications, there are multiple uncertain parameters in a model that must be estimated. However, high dimensionality of the parameters and computational complexity of the PDE solves make such problems challenging. A common approach is to reduce the dimension by fixing some parameters (which we will call auxiliary parameters) to a best estimate and use techniques from PDE-constrained optimization to estimate the other parameters. In this article, hyper-differential sensitivity analysis (HDSA) is used to assess the sensitivity of the solution of the PDE-constrained optimization problem to changes in the auxiliary parameters. Foundational assumptions for HDSA require satisfaction of the optimality conditions which are not always practically feasible as a result of ill-posedness in the inverse problem. Here we introduce novel theoretical and computational approaches to justify and enable HDSA for ill-posed inverse problems by projecting the sensitivities on likelihood informed subspaces and defining a posteriori updates. Our proposed framework is demonstrated on a nonlinear multiphysics inverse problem motivated by estimation of spatially heterogeneous material properties in the presence of spatially distributed parametric modeling uncertainties.
The sheer mass and nature of contaminated materials at DOE and DoD sites, makes it impractical to completely restore these sites to predisposal conditions. DOE faces long-term challenges, particularly with developing monitoring and end state approaches for clean-up that are protective of the environment, technically based and documented, sustainable, and most importantly cost effective. Integrated systems-based monitoring approaches (e.g., tools for characterization and monitoring, multi-component strategies, geophysical modeling) could provide novel approaches and a framework to (a) define risk-informed endpoints and/or conditions that constitute completion of cleanup and (b) provide the understanding for implementation of advanced scientific approaches to meet cleanup goals. Multi-component strategies which combine site conceptual models, biological, chemical, and physical remediation strategies, as well as iterative review and optimization have proven successful at several DOE sites. Novel tools such as enzyme probes and quantitative PCR for DNA and RNA, and innovative modeling approaches for complex subsurface environments, have been successful at facilitating the reduced operation or shutdown of pump and treat facilities and transition of clean-up activities into monitored natural attenuation remedies. Integrating novel tools with site conceptual models and other lines of evidence to characterize, optimize, and monitor long term remedial approaches for complex contaminant plumes are critical for transitioning active remediation into cost effective, yet technically defensible end pointstrategies.
This document describes progress in the ARPA-E DIFFERENTIATE project titled “Context-Aware Learning for Inverse Design in Photovoltaics” during the period of May 2019 to May 2022. This project is being performed at Iowa State University, New York University, Stanford University, and National Renewable Energy Laboratory. The project aims to develop a new machine learning (ML) framework to significantly accelerate the design of organic microstructures for improved organic photovoltaic performance. In this project, we had developed an inverse design framework using Deep Learning called InvNets for generating microstructures with desired physics-driven properties. As a preliminary product, in Milestone 3, we demonstrated how InvNets show 20% improvement in the performance of the microstructures and over 100X speedup in the performance compared to traditional processes for physics-driven inverse design. Later, in Milestone 6, we demonstrated that InvNets work for more complex physics properties, specifically, generating microstructures for organic photovoltaic cells with desired current-voltage characteristics. Further, in Milestone 4, we explored the idea of using physics-aware surrogates for obtaining solutions of partial differential equations(PDE) called as DiffNets(now called as NeuFENets to avoid ambiguity of names). The connection between both frameworks is that DiffNet surrogates form the physics-aware surrogate in the InvNet framework. Finally in Milestone 8, we extend our framework for other physics domains. Specifically, we explore building geometry-aware NeuFENets by developing physics surrogates that exploit ideas from traditional immersed boundary finite element methods. With these updates, we are able to achieve all the Milestones.
It is challenging to formulate complex physical phenomena that occur in a manufacturing process, particularly when the available data are limited, rendering conventional data-driven approaches ineffective. This study aims to predict humping onset in high-speed laser welding by introducing a novel framework, namely text-to-equations generative pre-trained transformer (T2EGPT). This method leverages the capabilities of large language models (LLMs), in combination with sparse experimental data and enriched literature data, to derive an interpretable and generalizable equation for predicting humping initiation. By capturing key correlations among physical parameters, T2EGPT generates a compact and dimensionless expression that accurately predicts hump formation. The equation reveals that humping arises from the interplay between inertia-driven backward melt flow and capillary-driven surface stabilization, where inertial forces drive molten metal backward and capillary forces resist surface deformation. Furthermore, compared to traditional data-driven models, T2EGPT demonstrates enhanced predictive accuracy and cross-material transferability. More broadly, this study highlights the potential of LLMs to integrate textual information with data-driven discovery, enabling the extraction of physical laws in data-scarce scientific domains.
3D printing originally known as additive manufacturing is a process of making 3 dimensional solid objects from a CAD file. This ground breaking technology is widely used for industrial and biomedical purposes such as building objects, tools, body parts and cosmetics. An important benefit of 3D printing is the cost reduction and manufacturing flexibility; complex parts are built at the fraction of the price. However, layer by layer printing of complex shapes adds error due to the surface roughness. Any such error results in poor quality products with inaccurate dimensions. The main purpose of this research is to measure the amount of printing errors for parts with different geometric shapes and to analyze them for finding optimal printing settings to minimize the error. We use a Design of Experiments framework, and focus on studying parts with cone and ellipsoid shapes. We found that the orientation and the shape of geometric shapes have significant effect on the printing error. From our analysis, we also determined the optimal orientation that gives the least printing error.
The capture of CO2 and its simultaneously conversion to useful chemical fuels driven by solar energy represents one of the best solutions to resolve our growing energy and environmental concerns. The most critical challenge to this endeavor is the rational design of a photocatalytic architecture that can effectively couple a given photosensitizer (PS) with an appropriate catalyst, thereby enabling efficient photosensitization of a multi-electron reduction catalysis. This research program aims to address this challenge using an interdisciplinary approach that combines innovative material design and synthesis, fundamental mechanistic studies, and photocatalytic performance evaluation. The strategies include 1) constructing and investigating a novel class of 2D COF hybrid photocatalysts with an effective photoactive organic building block as PS and a precisely incorporated CO2 reduction molecular catalyst (MC); and 2) mechanistic origins of CO2 photoreduction using a set of complementary time-resolved and in situ spectroscopic techniques. The novelty of the proposed hybrid system lies in the unprecedented combination of the unique advantage of porous crystalline COF PS with the precise catalytic function of MC for photocatalytic CO2 reduction. In the periods of the support (09/01/2019-12/31/2022), we have made research progress in four projects: 1) Exploring 2D COFs with incorporated Mn complex for light driven CO2 reduction; 2) The dependence of excited state and charge transfer dynamics on monomer structure of 2D COFs; and 3) Control over Charge Separation by Imine Structural Isomerization in Covalent Organic Frameworks with Implications on CO2 Photoreduction; and 4) The impact of monomer structure on the photoluminescence properties of COFs. We found that both monomer structure and linker chemistry can effectively impact the excited state dynamics, charge transfer properties, and photoluminescence quantum yields, the important properties that dictate their applications in photocatalysis. In addition, we found that the direction of imine linker determines charge transfer direction and thus controls the types of catalytic reactions (e.g. water oxidation or CO2 reduction reactions). The result from these fundamental studies provides important information for correlating the structure of the COF photocatalysts with their photophysical properties and catalytic functions, paving the way for their novel application in photocatalysis. We expect that our findings will contribute to addressing current shortcomings of semiconductor- and molecular-based photocatalytic systems that suffer from inefficient light harvesting and charge separation and poor adsorption and activation of reactants. In turn, this research will contribute towards the development of novel photocatalytic systems for CO2 reduction to generate renewable chemical fuels and simultaneously address the problem of mitigating climate change due to CO2 accumulation. In addition, the experimental approaches employed in this research can be easily transferred to other energy technologies and are expected to broadly impact fields involving photocatalysis, optoelectronic devices, and solar energy conversion. The proposed research has also been integrated with educational activities and serve as a basis to raise awareness around the critical issues of global energy production and consumption, and to develop the next generation of solar energy scientists.
Conspectus The variability of chemical bonding in open-shell transition-metal complexes not only motivates their study as functional materials and catalysts but also challenges conventional computational modeling tools. Here, tailoring ligand chemistry can alter preferred spin or oxidation states as well as electronic structure properties and reactivity, creating vast regions of chemical space to explore when designing new materials atom by atom. Although first-principles density functional theory (DFT) remains the workhorse of computational chemistry in mechanism deduction and property prediction, it is of limited use here. DFT is both far too computationally costly for widespread exploration of transition-metal chemical space and also prone to inaccuracies that limit its predictive performance for localized d electrons in transition-metal complexes. These challenges starkly contrast with the well-trodden regions of small-organic-molecule chemical space, where the analytical forms of molecular mechanics force fields and semiempirical theories have for decades accelerated the discovery of new molecules, accurate DFT functional performance has been demonstrated, and gold-standard methods from correlated wavefunction theory can predict experimental results to chemical accuracy. The combined promise of transition-metal chemical space exploration and lack of established tools has mandated a distinct approach. In this Account, we outline the path we charted in exploration of transition-metal chemical space starting from the first machine learning (ML) models (i.e., artificial neural network and kernel ridge regression) and representations for the prediction of open-shell transition-metal complex properties. The distinct importance of the immediate coordination environment of the metal center as well as the lack of low-level methods to accurately predict structural properties in this coordination environment first motivated and then benefited from these ML models and representations. Once developed, the recipe for prediction of geometric, spin state, and redox potential properties was straightforwardly extended to a diverse range of other properties, including in catalysis, computational “feasibility”, and the gas separation properties of periodic metal–organic frameworks. Interpretation of selected features most important for model prediction revealed new ways to encapsulate design rules and confirmed that models were robustly mapping essential structure–property relationships. Encountering the special challenge of ensuring that good model performance could generalize to new discovery targets motivated investigation of how to best carry out model uncertainty quantification. Distance-based approaches, whether in model latent space or in carefully engineered feature space, provided intuitive measures of the domain of applicability. With all of these pieces together, ML can be harnessed as an engine to tackle the large-scale exploration of transition-metal chemical space needed to satisfy multiple objectives using efficient global optimization methods. In practical terms, bringing these artificial intelligence tools to bear on the problems of transition-metal chemical space exploration has resulted in ML-model assessments of large, multimillion compound spaces in minutes and validated new design leads in weeks instead of decades.