Classification of battery compounds using structure-free Mendeleev encodings.

Machine learning is a valuable tool that can accelerate the discovery and design of materials occupying combinatorial chemical spaces. However, the prerequisite need for vast amounts of training data can be prohibitive when significant resources are needed to characterize or simulate candidate structures. Recent results have shown that structure-free encoding of complex materials, based entirely on chemical compositions, can overcome this impediment and perform well in unsupervised learning tasks. In this study, we extend this exploration to supervised classification, and show how structure-free encoding can accurately predict classes of material compounds for battery applications without time consuming measurement of bonding networks, lattices or densities. SCIENTIFIC CONTRIBUTION: The comprehensive evaluation of structure-free encodings of complex materials in classification tasks, including binary and multi-class separation, inclusive of three classifiers based on different logic function, is measured four metrics and learning curves. The encoding is applied to two data sets from computational and experimental sources, and the outcomes visualised using 5 approaches to confirms the suitability and superiority of Mendeleev encoding. These methods are general and accessible using source software, to provide simple, intuitive and interpretable materials informatics outcomes to accelerate materials design.

Complex Dispersion of Detonation Nanodiamond Revealed by Machine Learning Assisted Cryo-TEM and Coarse-Grained Molecular Dynamics Simulations.

Understanding the polydispersity of nanoparticles is crucial for establishing the efficacy and safety of their role as drug delivery carriers in biomedical applications. Detonation nanodiamonds (DNDs), 3-5 nm diamond nanoparticles synthesized through detonation process, have attracted great interest for drug delivery due to their colloidal stability in water and their biocompatibility. More recent studies have challenged the consensus that DNDs are monodispersed after their fabrication, with their aggregate formation poorly understood. Here, we present a novel characterization method of combining machine learning with direct cryo-transmission electron microscopy imaging to characterize the unique colloidal behavior of DNDs. Together with small-angle X-ray scattering and mesoscale simulations we show and explain the clear differences in the aggregation behavior between positively and negatively charged DNDs. Our new method can be applied to other complex particle systems, which builds essential knowledge for the safe implementation of nanoparticles in drug delivery.

Explainable discovery of disease biomarkers: The case of ovarian cancer to illustrate the best practice in machine learning and Shapley analysis.

OBJECTIVE: Ovarian cancer is a significant health issue with lasting impacts on the community. Despite recent advances in surgical, chemotherapeutic and radiotherapeutic interventions, they have had only marginal impacts due to an inability to identify biomarkers at an early stage. Biomarker discovery is challenging, yet essential for improving drug discovery and clinical care. Machine learning (ML) techniques are invaluable for recognising complex patterns in biomarkers compared to conventional methods, yet they can lack physical insights into diagnosis. eXplainable Artificial Intelligence (XAI) is capable of providing deeper insights into the decision-making of complex ML algorithms increasing their applicability. We aim to introduce best practice for combining ML and XAI techniques for biomarker validation tasks. METHODS: We focused on classification tasks and a game theoretic approach based on Shapley values to build and evaluate models and visualise results. We described the workflow and apply the pipeline in a case study using the CDAS PLCO Ovarian Biomarkers dataset to demonstrate the potential for accuracy and utility. RESULTS: The case study results demonstrate the efficacy of the ML pipeline, its consistency, and advantages compared to conventional statistical approaches. CONCLUSION: The resulting guidelines provide a general framework for practical application of XAI in medical research that can inform clinicians and validate and explain cancer biomarkers.

Unsupervised machine learning discovers classes in aluminium alloys.

Aluminium (Al) alloys are critical to many applications. Although Al alloys have been commercially widespread for over a century, their development has predominantly taken a trial-and-error approach. Furthermore, many discrete studies regarding Al alloys, often application specific, have precluded a broader consolidation of Al alloy classification. Iterative label spreading (ILS), an unsupervised machine learning approach, was used to identify the different classes of Al alloys, drawing from a specifically curated dataset of 1154 Al alloys (including alloy composition and processing conditions). Using ILS, eight classes of Al alloys were identified based on a comprehensive feature set under two descriptors. Further, a decision tree classifier was used to validate the separation of classes.

Liquid metal synthesis solvents for metallic crystals.

In nature, snowflake ice crystals arrange themselves into diverse symmetrical six-sided structures. We show an analogy of this when zinc (Zn) dissolves and crystallizes in liquid gallium (Ga). The low-melting-temperature Ga is used as a "metallic solvent" to synthesize a range of flake-like Zn crystals. We extract these metallic crystals from the liquid metal solvent by reducing its surface tension using a combination of electrocapillary modulation and vacuum filtration. The liquid metal-grown crystals feature high morphological diversity and persistent symmetry. The concept is expanded to other single and binary metal solutes and Ga-based solvents, with the growth mechanisms elucidated through ab initio simulation of interfacial stability. This strategy offers general routes for creating highly crystalline, shape-controlled metallic or multimetallic fine structures from liquid metal solvents.

Safety-by-design using forward and inverse multi-target machine learning.

The economic and social future of nanotechnology depends on our ability and manufacture nanomaterials that avoid potential toxicity, by identifying them before they are made, used and released into the environment. Safety-by-design is a framework for including these issues at an early stage of the development process, but balancing multiple nanoparticle properties and selection criteria remains challenging. Based on a synthetic data set of over 19,000 possible sunscreen product specifications, we have used multi-target machine learning to predict the corresponding size, shape, concentration and polytype of titania nanoparticle additives. The study considers the optical properties responsible for the sun protection factor and product transparency, including the extinction coefficients for ultra violet and visible light, and the potential for toxicity due to the generation of reactive oxygen species from the photocatalytically active facets of both anatase and rutile nanoparticles, as a function of the size and shape. We predict a number of conventional forward structure/property and property/product relationships, but show that a direct structure/product relationship provides superior performance when predicting multiple properties or product specifications simultaneously. These models are then inverted, re-optimized and re-trained to provide focused, high performing inverse design models that do not require additional optimization, and are capable of identifying nanoparticle configurations outside of the training set. The ability to directly predict suitable nanoparticle structures that conform to prerequisite sun protection, transparently and potential toxicity thresholds represents a new approach to safety-by-design that can be applied to other products and materials where multiple design criteria must be met at the same time.

Impact of atomistic or crystallographic descriptors for classification of gold nanoparticles.

Machine learning models are known to be sensitive to the features used to train them, but there is currently no way to predict the impact of using different features prior to feature extraction. This is particularly important to fields such as nanotechnology that are highly multi-disciplinary, and samples can be characterised many different ways depending on the preferences of individual researchers. Does it matter if nanomaterials are described using the interatomic coordinations or more complex order parameters? In this study we compare results of supervised and unsupervised learning on a single set of gold nanoparticles that has been characterised by two different descriptors, each with a unique feature space. We find that there are some consistencies, and model selection is descriptor-agnostic, but the level of detail and the type of information that can be extracted from the results is sensitive to the way the particles are described. Unsupervised clustering revealed that an atomistic descriptor provides a finer-grained interpretation and clusters that are sub-clusters of a more sophisticated crystallographic descriptor, which is consistent with both how the features were calculated, and how they are interpreted in the domain. A supervised classifier revealed that the types of features responsible for the separation are related to the bulk structure, regardless of the descriptor, but capture different types of information. For both the atomistic and crystallographic descriptor the gradient boosting decision tree classifier gave superior results of F1-scores of 0.96 and 0.98, respectively, with excellent precision and recall, even though the clustering presented a challenging multi-classification problem.

Enhancing classical gold nanoparticle simulations with electronic corrections and machine learning.

Classical simulations of materials and nanoparticles have the advantage of speed and scalability but at the cost of precision and electronic properties, while electronic structure simulations have the advantage of accuracy and transferability but are typically limited to small and simple systems due to the increased computational complexity. Machine learning can be used to bridge this gap by providing correction terms that deliver electronic structure results based on classical simulations, to retain the best of both worlds. In this study we train an artificial neural network (ANN) as a general ansatz to predict a correction of the total energy of arbitrary gold nanoparticles based on general (material agnostic) features, and a limited set of structures simulated with an embedded atom potential and the self-consistent charge density functional tight binding method. We find that an accurate model with an overall precision of 14 eV or 8.6% can be found using a diverse range of particles and a large number of manually generated features which were then reduced using automatic data-driven approach to reduce evaluation bias. We found the ANN reduces to a linear relationship if a suitable subset of important features are identified prior to training, and that the prediction can be improved by classifying the nanoparticles into kinetically limited and thermodynamically limited subsets based prior to training the ANN corrections. The results demonstrate the potential for machine learning to enhance classical molecular dynamics simulations without adding significant computational complexity, and provides methodology that could be used to predict other electronic properties which cannot be calculated solely using classical simulations.

Unsupervised structure classes vs. supervised property classes of silicon quantum dots using neural networks.

Machine learning classification is a useful technique to predict structure/property relationships in samples of nanomaterials where distributions of sizes and mixtures of shapes are persistent. The separation of classes, however, can either be supervised based on domain knowledge (human intelligence), or based entirely on unsupervised machine learning (artificial intelligence). This raises the questions as to which approach is more reliable, and how they compare? In this study we combine an ensemble data set of electronic structure simulations of the size, shape and peak wavelength for the optical emission of hydrogen passivated silicon quantum dots with artificial neural networks to explore the utility of different types of classes. By comparing the domain-driven and data-driven approaches we find there is a disconnect between what we see (optical emission) and assume (that a particular color band represents a special class), and what the data supports. Contrary to expectation, controlling a limited set of structural characteristics is not specific enough to classify a quantum dot based on color, even though it is experimentally intuitive.

The pure and representative types of disordered platinum nanoparticles from machine learning.

The development of interpretable structure/property relationships is a cornerstone of nanoscience, but can be challenging when the structural diversity and complexity exceeds our ability to characterise it. This is often the case for imperfect, disordered and amorphous nanoparticles, where even the nomenclature can be unspecific. Disordered platinum nanoparticles have exhibited superior performance for some reactions, which makes a systematic way of describing them highly desirable. In this study we have used a diverse set of disorder platinum nanoparticles and machine learning to identify the pure and representative structures based on their similarity in 121 dimensions. We identify two prototypes that are representative of separable classes, and seven archetypes that are the pure structures on the convex hull with which all other possibilities can be described. Together these nine nanoparticles can explain all of the variance in the set, and can be described as either single crystal, twinned, spherical or branched; with or without roughened surfaces. This forms a robust sub-set of platinum nanoparticle upon which to base further work, and provides a theoretical basis for discussing structure/property relationships of platinum nanoparticles that are not geometrically ideal.

Simulating facet-dependent aggregation and assembly of distributions of polyhedral nanoparticles.

Coarse-grained molecular dynamics simulations of diamond nanoparticles were performed to investigate the effects of size polydispersity on three polyhedral shapes chosen to span a diverse space of surface interactions. It was found that the resulting self-assembly was size dependent as the simulations were quenched, with the largest nanoparticles providing a clustered scaffold for subsequent smaller nanoparticle assembly. Additionally, facet-facet interactions were dominated by the {111} surface and the resulting aggregate was dominated by meso-sized porosity for monodisperse systems, broadening to larger diameters for polydisperse systems.

Machine learning reveals multiple classes of diamond nanoparticles.

Generating samples of nanoparticles with specific properties that allow for structural diversity, rather than requiring structural precision, is a more sustainable prospect for industry, where samples need to be both targeted to specific applications and cost effective. This can be better enabled by defining classes of nanoparticles and characterising the properties of the class as a whole. In this study, we use machine learning to predict the different classes of diamond nanoparticles based entirely on the structural features and explore the populations of these classes in terms of the size, shape, speciation and charge transfer properties. We identify 9 different types of diamond nanoparticles based on their similarity in 17 dimensions and, contrary to conventional wisdom, find that the fraction of sp2 or sp3 hybridized atoms are not strong determinants, and that the classes are only weakly related to size. Each class has been describe in such way as to enable rapid assignment using microanalysis techniques.

Dynamic self-assembly of detonation nanodiamond in water.

Nanodiamonds are increasingly used in many areas of science and technology, yet, their colloidal properties remain poorly understood. Here we use direct imaging as well as light and X-ray scattering reveal that purified detonation nanodiamond (DND) particles in an aqueous environment exhibit a self-assembled lace-like network, even without additional surface modification. Such behaviour is previously unknown and contradicts the current consensus that DND exists as mono-dispersed single particles. With the aid of mesoscale simulations, we show that the lace network is likely the result of competition between a short-ranged electrostatic attraction between faceted particles and a longer-ranged repulsion arising from the interaction between the surface functional groups and the surrounding water molecules which prevents complete flocculation. Our findings have significant implications for applications of DND where control of the aggregation behaviour is critical to performance.

Predicting archetypal nanoparticle shapes using a combination of thermodynamic theory and machine learning.

Machine learning is a useful way of identifying representative or pure nanoparticle shapes as part of a larger ensemble, but its predictive capabilities can be limited when a large dataset of candidate structures must already exist. Ideally one would like to use machine learning to define the ideal dataset for future, more computationally intensive, studies before a significant amount of resources are consumed. In this work we combine an established analytical phenomenological model and statistical machine learning to predict the archetypes and prototypes of a diverse ensemble of 2380 platinum nanoparticle morphologies developed with less than twenty input electronic structure simulations. By parameterising a size- and shape-dependent thermodynamic model, probabilities are assigned to seventeen different shapes between three and thirty nanometres, which together with structural features such as nanoparticle diameter, surface area, sphericity and facet configuration form the basis for archetypal analysis and K-means clustering. Using this approach we rapidly identify six "pure" archetypes and twelve "representative" prototypes that can be used in future computational studies of properties such as catalysis.

Correlating anisotropy and disorder with the surface structure of platinum nanoparticles.

Due to the competition between numerous physicochemical variables during formation and processing, platinum nanocatalysts typically contain a mixture of shapes, distributions of sizes, and a considerable degree of surface imperfection. Structural imperfection and sample polydispersivity are inevitable at scale, but accepting bulk and surface diversity as legitimate design features provides new opportunities for nanoparticle design. In recent years disorder and anisotropy have been embraced as useful design parameters but predicting the impact of uncontrollable imperfection a priori is challenging. In the present work we have created an ensemble of uniquely imperfect nanoparticles extracted from classical molecular dynamics trajectories and applied statistical filters to restrict the ensemble in ways that reflect common industrial design principles. We find that targeting different sizes and size distributions may be an effective way of promoting or suppressing internal disorder or crystallinity (as required), but the degree of anisotropy of the particle as a whole has a greater impact on the population of different types of surface ordering and active sites. These results indicate that tuning of disordered and anisotropic Pt nanoparticles is possible, but it is not as straightforward as geometrically ideal nanoparticles with a high degree of crystallinity. It is unlikely that a synthesis strategy could eliminate this diversity entirely, or ensure this type of structural complexity does not develop post-synthesis under operational conditions, but it may be possible to bias the formation of specific bulk structures and the surface anisotropy.

Three-Dimensional Branched and Faceted Gold-Ruthenium Nanoparticles: Using Nanostructure to Improve Stability in Oxygen Evolution Electrocatalysis.

Achieving stability with highly active Ru nanoparticles for electrocatalysis is a major challenge for the oxygen evolution reaction. As improved stability of Ru catalysts has been shown for bulk surfaces with low-index facets, there is an opportunity to incorporate these stable facets into Ru nanoparticles. Now, a new solution synthesis is presented in which hexagonal close-packed structured Ru is grown on Au to form nanoparticles with 3D branches. Exposing low-index facets on these 3D branches creates stable reaction kinetics to achieve high activity and the highest stability observed for Ru nanoparticle oxygen evolution reaction catalysts. These design principles provide a synthetic strategy to achieve stable and active electrocatalysts.

Size dependent surface reconstruction in detonation nanodiamonds.

Nanometer-sized diamond particles are used in bio-medical applications, where the nature of the nanodiamond surfaces is crucial to achieving correct functionalisation. Herein, using high-resolution transmission electron microscopy and electronic structure calculations, we study the surface reconstructions that occur in detonation-synthesized nanodiamonds. Our results show that particles smaller than 3 nm exhibit size- and shape-dependent surface reconstructions, and that the surfaces can exhibit a higher-than-expected fraction of sp2+x bonding. This indicates an aliphatic character for sub-3 nm nanodiamond particles. Such behaviour impacts the functionality of nanodiamonds, where both size and surface charge can drive performance. Our observations offer a potential strategy for better functionalization control via the size range of the particles.

Machine Learning for Silver Nanoparticle Electron Transfer Property Prediction.

Nanoparticles exhibit diverse structural and morphological features that are often interconnected, making the correlation of structure/property relationships challenging. In this study a multi-structure/single-property relationship of silver nanoparticles is developed for the energy of Fermi level, which can be tuned to improve the transfer of electrons in a variety of applications. By combining different machine learning analytical algorithms, including k-mean, logistic regression, and random forest with electronic structure simulations, we find that the degree of twinning (characterized by the fraction of hexagonal closed packed atoms) and the population of the {111} facet (characterized by a surface coordination number of nine) are strongly correlated to the Fermi energy of silver nanoparticles. A concise three layer artificial neural network together with principal component analysis is built to predict this property, with reduced geometrical, structural, and topological features, making the method ideal for efficient and accurate high-throughput screening of large-scale virtual nanoparticle libraries and the creation of single-structure/single-property, multi-structure/single-property, and single-structure/multi-property relationships in the near future.

The impact of size and shape distributions on the electron charge transfer properties of silver nanoparticles.

Many applications of silver nanoparticles are moderated by the electron charge transfer properties, such as the ionization potential, electron affinity and Fermi energy, which may be tuned by controlling the size and shape of individual particles. However, since producing samples of silver nanoparticles that are perfectly monodispersed in terms of both size and shape can be prohibitive, it is important to understand how these properties are impacted by polydispersivity, and ideally be able to predict the tolerance for variation of different geometric features. In this study, we use straightforward statistical methods, together with electronic structure simulations, to predict the electron charge transfer properties of different types of ensembles of silver nanoparticles and how restricting the structural diversity in different ways can improve or retard performance. In agreement with previous reports, we confirm that restricting the shape distribution will tune the charge transfer properties toward specific reactions, but by including the quality factors for each case we go beyond this assessment and show how targeting specific classes of morphologies and restricting the distribution of size can impact sensitivity.

Bias-Free Chemically Diverse Test Sets from Machine Learning.

Current benchmarking methods in quantum chemistry rely on databases that are built using a chemist's intuition. It is not fully understood how diverse or representative these databases truly are. Multivariate statistical techniques like archetypal analysis and K-means clustering have previously been used to summarize large sets of nanoparticles however molecules are more diverse and not as easily characterized by descriptors. In this work, we compare three sets of descriptors based on the one-, two-, and three-dimensional structure of a molecule. Using data from the NIST Computational Chemistry Comparison and Benchmark Database and machine learning techniques, we demonstrate the functional relationship between these structural descriptors and the electronic energy of molecules. Archetypes and prototypes found with topological or Coulomb matrix descriptors can be used to identify smaller, statistically significant test sets that better capture the diversity of chemical space. We apply this same method to find a diverse subset of organic molecules to demonstrate how the methods can easily be reapplied to individual research projects. Finally, we use our bias-free test sets to assess the performance of density functional theory and quantum Monte Carlo methods.