Determining the orderliness of carbon materials with nanoparticle imaging and explainable machine learning.

Carbon materials have paramount importance in various fields of materials science, from electronic devices to industrial catalysts. The properties of these materials are strongly related to the distribution of defects-irregularities in electron density on their surfaces. Different materials have various distributions and quantities of these defects, which can be imaged using a procedure that involves depositing palladium nanoparticles. The resulting scanning electron microscopy (SEM) images can be characterized by a key descriptor-the ordering of nanoparticle positions. This work presents a highly interpretable machine learning approach for distinguishing between materials with ordered and disordered arrangements of defects marked by nanoparticle attachment. The influence of the degree of ordering was experimentally evaluated on the example of catalysis via chemical reactions involving carbon-carbon bond formation. This represents an important step toward automated analysis of SEM images in materials science.

Quantitative Determination of Active Species Transforming the R-NHC Coupling Process under Catalytic Conditions.

Palladium complexes with N-heterocyclic carbenes (Pd/NHC) serve as prominent precatalysts in numerous Pd-catalyzed organic reactions. While the evolution of Pd/NHC complexes, which involves the cleavage of the Pd-C(NHC) bond via reductive elimination and dissociation, is acknowledged to influence the catalysis mechanism and the performance of the catalytic systems, conventional analytic techniques [such as NMR, IR, UV-vis, gas chromatography-mass spectrometry (GC-MS), and high-performance liquid chromatography (HPLC)] frequently fail to quantitatively monitor the transformations of Pd/NHC complexes at catalyst concentrations typical of real-world conditions (below approximately 1 mol %). In this study, for the first time, we show the viability of using electrospray ionization mass spectrometry (ESI-MS). This approach was combined with the use of selectively deuterated H-NHC, Ph-NHC, and O-NHC coupling products as internal standards, allowing for an in-depth quantitative analysis of the evolution of Pd/NHC catalysts within actual catalytic systems. The reliability of this approach was affirmed by aligning the ESI-MS results with the NMR spectroscopy data obtained at greater Pd/NHC precatalyst concentrations (2-5 mol %) in the Mizoroki-Heck, Sonogashira, and alkyne transfer hydrogenation reactions. The efficacy of the ESI-MS methodology was further demonstrated through its application in the Mizoroki-Heck reaction at Pd/NHC loadings of 5, 0.5, 0.05, and 0.005 mol %. In this work, for the first time, we present a methodology for the quantitative characterization of pivotal catalyst transformation processes commonly observed in M/NHC systems.

Autonomous chemical research with large language models.

Transformer-based large language models are making significant strides in various fields, such as natural language processing1-5, biology6,7, chemistry8-10 and computer programming11,12. Here, we show the development and capabilities of Coscientist, an artificial intelligence system driven by GPT-4 that autonomously designs, plans and performs complex experiments by incorporating large language models empowered by tools such as internet and documentation search, code execution and experimental automation. Coscientist showcases its potential for accelerating research across six diverse tasks, including the successful reaction optimization of palladium-catalysed cross-couplings, while exhibiting advanced capabilities for (semi-)autonomous experimental design and execution. Our findings demonstrate the versatility, efficacy and explainability of artificial intelligence systems like Coscientist in advancing research.

Time-Resolved Formation and Operation Maps of Pd Catalysts Suggest a Key Role of Single Atom Centers in Cross-Coupling.

An approach to the spatially localized characterization of supported catalysts over a reaction course is proposed. It consists of a combination of scanning, transmission, and high-resolution scanning transmission electron microscopy to determine metal particles from arrays of surface nanoparticles to individual nanoparticles and individual atoms. The study of the evolution of specific metal catalyst particles at different scale levels over time, particularly before and after the cross-coupling catalytic reaction, made it possible to approach the concept of 4D catalysis-tracking the positions of catalytic centers in space (3D) over time (+1D). The dynamic behavior of individual palladium atoms and nanoparticles in cross-coupling reactions was recorded with nanometer accuracy via the precise localization of catalytic centers. Single atoms of palladium leach out into solution from the support under the action of the catalytic system, where they exhibit extremely high catalytic activity compared to surface metal nanoparticles. Monoatomic centers, which make up only approximately 1% of palladium in the Pd/C system, provide more than 99% of the catalytic activity. The remaining palladium nanoparticles changed their shape and could move over the surface of the support, which was recorded by processing images of the array of nanoparticles with a neural network and aligning them using automatically detected keypoints. The study reveals a novel opportunity for single-atom catalysis─easier detachment (capture) from (on) the carbon support surface is the origin of superior catalytic activity, rather than the operation of single atomic catalytic centers on the surface of the support, as is typically assumed.

Automated Recognition of Nanoparticles in Electron Microscopy Images of Nanoscale Palladium Catalysts.

Automated computational analysis of nanoparticles is the key approach urgently required to achieve further progress in catalysis, the development of new nanoscale materials, and applications. Analysis of nanoscale objects on the surface relies heavily on scanning electron microscopy (SEM) as the experimental analytic method, allowing direct observation of nanoscale structures and morphology. One of the important examples of such objects is palladium on carbon catalysts, allowing access to various chemical reactions in laboratories and industry. SEM images of Pd/C catalysts show a large number of nanoparticles that are usually analyzed manually. Manual analysis of a statistically significant number of nanoparticles is a tedious and highly time-consuming task that is impossible to perform in a reasonable amount of time for practically needed large amounts of samples. This work provides a comprehensive comparison of various computer vision methods for the detection of metal nanoparticles. In addition, multiple new types of data representations were developed, and their applicability in practice was assessed.

Fully Automated Unconstrained Analysis of High-Resolution Mass Spectrometry Data with Machine Learning.

Mass spectrometry (MS) is a convenient, highly sensitive, and reliable method for the analysis of complex mixtures, which is vital for materials science, life sciences fields such as metabolomics and proteomics, and mechanistic research in chemistry. Although it is one of the most powerful methods for individual compound detection, complete signal assignment in complex mixtures is still a great challenge. The unconstrained formula-generating algorithm, covering the entire spectra and revealing components, is a "dream tool" for researchers. We present the framework for efficient MS data interpretation, describing a novel approach for detailed analysis based on deisotoping performed by gradient-boosted decision trees and a neural network that generates molecular formulas from the fine isotopic structure, approaching the long-standing inverse spectral problem. The methods were successfully tested on three examples: fragment ion analysis in protein sequencing for proteomics, analysis of the natural samples for life sciences, and study of the cross-coupling catalytic system for chemistry.

Integration of thermal imaging and neural networks for mechanical strength analysis and fracture prediction in 3D-printed plastic parts.

Additive manufacturing demonstrates tremendous progress and is expected to play an important role in the creation of construction materials and final products. Contactless (remote) mechanical testing of the materials and 3D printed parts is a critical limitation since the amount of collected data and corresponding structure/strength correlations need to be acquired. In this work, an efficient approach for coupling mechanical tests with thermographic analysis is described. Experiments were performed to find relationships between mechanical and thermographic data. Mechanical tests of 3D-printed samples were carried out on a universal testing machine, and the fixation of thermal changes during testing was performed with a thermal imaging camera. As a proof of concept for the use of machine learning as a method for data analysis, a neural network for fracture prediction was constructed. Analysis of the measured data led to the development of thermographic markers to enhance the thermal properties of the materials. A combination of artificial intelligence with contactless nondestructive thermal analysis opens new opportunities for the remote supervision of materials and constructions.

Toward Totally Defined Nanocatalysis: Deep Learning Reveals the Extraordinary Activity of Single Pd/C Particles.

Homogeneous catalysis is typically considered "well-defined" from the standpoint of catalyst structure unambiguity. In contrast, heterogeneous nanocatalysis often falls into the realm of "poorly defined" systems. Supported catalysts are difficult to characterize due to their heterogeneity, variety of morphologies, and large size at the nanoscale. Furthermore, an assortment of active metal nanoparticles examined on the support are negligible compared to those in the bulk catalyst used. To solve these challenges, we studied individual particles of the supported catalyst. We made a significant step forward to fully characterize individual catalyst particles. Combining a nanomanipulation technique inside a field-emission scanning electron microscope with neural network analysis of selected individual particles unexpectedly revealed important aspects of activity for widespread and commercially important Pd/C catalysts. The proposed approach unleashed an unprecedented turnover number of 109 attributed to individual palladium on a nanoglobular carbon particle. Offered in the present study is the Totally Defined Catalysis concept that has tremendous potential for the mechanistic research and development of high-performance catalysts.

Deep neural network analysis of nanoparticle ordering to identify defects in layered carbon materials.

Smoothness/defectiveness of the carbon material surface is a key issue for many applications, spanning from electronics to reinforced materials, adsorbents and catalysis. Several surface defects cannot be observed with conventional analytic techniques, thus requiring the development of a new imaging approach. Here, we evaluate a convenient method for mapping such "hidden" defects on the surface of carbon materials using 1-5 nm metal nanoparticles as markers. A direct relationship between the presence of defects and the ordering of nanoparticles was studied experimentally and modeled using quantum chemistry calculations and Monte Carlo simulations. An automated pipeline for analyzing microscopic images is described: the degree of smoothness of experimental images was determined by a classification neural network, and then the images were searched for specific types of defects using a segmentation neural network. An informative set of features was generated from both networks: high-dimensional embeddings of image patches and statics of defect distribution.

Neural Network Analysis of Electron Microscopy Video Data Reveals the Temperature-Driven Microphase Dynamics in the Ions/Water System.

Real-time field-emission scanning electron microscopy (FE-SEM) measurements and neural network analysis were successfully merged to observe the temperature-induced behavior of soft liquid microdomains in mixtures of different ionic liquids with water. The combination of liquid FE-SEM and in situ heating techniques revealed temperature-driven solution restructuring for ions/water systems with different water states and their critical point behavior expressed in a rapid switch between thermal expansion and shrinkage of liquid microphases at temperatures of ≈100-130 °C, which was directly recorded on electron microscopy videos. Automation of FE-SEM video analysis by a neural network approach allowed quantification of the morphological changes in ions/water systems during heating on the basis of thousands of images processed with a speed almost equal to the frame rate of original electron microscopy videos. Tracking and evolution of the micro-heterogeneous domains, hypothesized in the Ioliomics concept, was mapped and quantified for the first time. The present study describes the concept for quick acquisition of big data in electron microscopy, develops rapid neural network analysis and shows how to link microscopic data to fundamental molecular properties.

Mechanistic Study of Pd/NHC-Catalyzed Sonogashira Reaction: Discovery of NHC-Ethynyl Coupling Process.

The product of a revealed transformation-NHC-ethynyl coupling-was observed as a catalyst transformation pathway in the Sonogashira cross-coupling, catalyzed by Pd/NHC complexes. The 2-ethynylated azolium salt was isolated in individual form and fully characterized, including X-ray analysis. A number of possible intermediates of this transformation with common formulae (NHC)n Pd(C2 Ph) (n=1,2) were observed and subjected to collision-induced dissociation (CID) and infrared multiphoton dissociation (IRMPD) experiments to elucidate their structure. Measured bond dissociation energies (BDEs) and IRMPD spectra were in an excellent agreement with quantum calculations for coupling product π-complexes with Pd0 . Molecular dynamics simulations confirmed the observed multiple CID fragmentation pathways. An unconventional methodology to study catalyst evolution suggests the reported transformation to be considered in the development of new catalytic systems for alkyne functionalization reactions.

Electron microscopy dataset for the recognition of nanoscale ordering effects and location of nanoparticles.

A unique ordering effect has been observed in functional catalytic nanoscale materials. Instead of randomly arranged binding to the catalyst surface, metal nanoparticles show spatially ordered behavior resulting in formation of geometrical patterns. Understanding of such nanoscale materials and analysis of corresponding microscopy images will never be comprehensive without appropriate reference datasets. Here we describe the first dataset of electron microscopy images comprising individual nanoparticles which undergo ordering on a surface towards the formation of geometrical patterns. The dataset developed in this study spans three levels of nanoscale organization: (i) individual nanoparticles (1-5 nm) and arrays of nanoparticles (5-20 nm), (ii) ordering effects (20-200 nm) and (iii) complex patterns (from nm to µm scales). The described dataset for the first time provides a possibility for the development of machine learning algorithms to study the unique phenomena of nanoparticles ordering and hierarchical organization.