Retrieving the Quantitative Chemical Information at Nanoscale from Scanning Electron Microscope Energy Dispersive X-ray Measurements by Machine Learning

Non-negative matrix factorization-aided phase unmixing and trace element quantification of STEM-EDXS data

Energy-dispersive X-ray spectroscopy (EDXS) mapping with a scanning transmission electron microscope (STEM) is commonly used for chemical characterization of materials. However, STEM-EDXS quantification becomes challenging when the phases constituting the sample under investigation share common elements and overlap spatially. In this paper, we present a methodology to identify, segment, and unmix phases with a substantial spectral and spatial overlap in a semi-automated fashion through combining non-negative matrix factorization with a priori knowledge of the sample. We illustrate the methodology using a sample taken from an electron beam-sensitive mineral assemblage representing Earth's deep mantle. With it, we retrieve the true EDX spectra of the constituent phases and their corresponding phase abundance maps. It further enables us to achieve a reliable quantification for trace elements having concentration levels of ∼100 ppm. Our approach can be adapted to aid the analysis of many materials systems that produce STEM-EDXS datasets having phase overlap and/or limited signal-to-noise ratio (SNR) in spatially-integrated spectra.

espm: A Python library for the simulation of STEM-EDXS datasets

We present two open-source Python packages: "electron spectro-microscopy" (espm) and "electron microscopy tables" (emtables). The espm software enables the simulation of scanning transmission electron microscopy energy-dispersive X-ray spectroscopy datacubes, based on user-defined chemical compositions and spatial abundance maps of constituent phases. The simulation process uses X-ray emission cross-sections generated via state-of-the-art calculations made with emtables. These tables are designed to be easily modifiable, either manually or using espm. The simulation framework is designed to test the application of decomposition algorithms for the analysis of STEM-EDX spectrum images with access to a known ground truth. We validate our approach using the case of a complex geology-related sample, comparing raw simulated and experimental datasets and the outputs of their non-negative matrix factorization. In addition to testing machine learning algorithms, our packages will also help experimental design, for instance, predicting dataset characteristics or establishing minimum counts needed to measure nanoscale features.

Two-Dimensional Polymers and Polymerizations

Synthetic chemists have developed robust methods to synthesize discrete molecules, linear and branched polymers, and disordered cross-linked networks. However, two-dimensional polymers (2DPs) prepared from designed monomers have been long missing from these capabilities, both as objects of chemical synthesis and in nature. Recently, new polymerization strategies and characterization methods have enabled the unambiguous realization of covalently linked macromolecular sheets. Here we review 2DPs and 2D polymerization methods. Three predominant 2D polymerization strategies have emerged to date, which produce 2DPs either as monolayers or multilayer assemblies. We discuss the fundamental understanding and scope of each of these approaches, including: the bond-forming reactions used, the synthetic diversity of 2DPs prepared, their multilayer stacking behaviors, nanoscale and mesoscale structures, and macroscale morphologies. Additionally, we describe the analytical tools currently available to characterize 2DPs in their various isolated forms. Finally, we review emergent 2DP properties and the potential applications of planar macromolecules. Throughout, we highlight achievements in 2D polymerization and identify opportunities for continued study.

Spectroscopic and Spectrometric Approaches for Assessing the Composition of Embedded Metals in Tissues

Many medical devices contain metals that interface with the body. Additionally, embedded metal fragments from military wounds are typically not removed, to avoid the risk of morbidity associated with invasive surgery. The long-term health consequences of many of these materials are not thoroughly understood. To this end, we have exposed rats for up to one year to implanted single-element metal pellets of any one of Al, Co, Cu, Fe, Ni, Pb, Ta, or W. Various tissues were harvested and flash frozen for analysis of their metal distribution. We discuss approaches to most thoroughly and reliably evaluate the distribution of metal in these tissues. The path to the most appropriate analytical technique took us through extensive examination of the tissues using scanning electron microscopy with energy dispersive X-ray spectroscopy (XPS), X-ray photoelectron spectroscopy (XPS), and laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS). Though any one of these methods is highly relied upon in surface chemistry analysis, LA-ICP-MS alone showed presence of metal in the tissue. This information will help build robust methods to bridge the gap in our understanding of biosolubility and distribution of embedded metal throughout the body.

Directly Probing the Local Coordination, Charge State, and Stability of Single Atom Catalysts by Advanced Electron Microscopy: A Review

The drive for atom efficient catalysts with carefully controlled properties has motivated the development of single atom catalysts (SACs), aided by a variety of synthetic methods, characterization techniques, and computational modeling. The distinct capabilities of SACs for oxidation, hydrogenation, and electrocatalytic reactions have led to the optimization of activity and selectivity through composition variation. However, characterization methods such as infrared and X-ray spectroscopy are incapable of direct observations at atomic scale. Advances in transmission electron microscopy (TEM) including aberration correction, monochromators, environmental TEM, and micro-electro-mechanical system based in situ holders have improved catalysis study, allowing researchers to peer into regimes previously unavailable, observing critical structural and chemical information at atomic scale. This review presents recent development and applications of TEM techniques to garner information about the location, bonding characteristics, homogeneity, and stability of SACs. Aberration corrected TEM imaging routinely achieves sub-Ångstrom resolution to reveal the atomic structure of materials. TEM spectroscopy provides complementary information about local composition, chemical bonding, electronic properties, and atomic/molecular vibration with superior spatial resolution. In situ/operando TEM directly observe the evolution of SACs under reaction conditions. This review concludes with remarks on the challenges and opportunities for further development of TEM to study SACs.

Charting the low-loss region in electron energy loss spectroscopy with machine learning

Exploiting the information provided by electron energy-loss spectroscopy (EELS) requires reliable access to the low-loss region where the zero-loss peak (ZLP) often overwhelms the contributions associated to inelastic scatterings off the specimen. Here we deploy machine learning techniques developed in particle physics to realise a model-independent, multidimensional determination of the ZLP with a faithful uncertainty estimate. This novel method is then applied to subtract the ZLP for EEL spectra acquired in flower-like WS₂ nanostructures characterised by a 2H/3R mixed polytypism. From the resulting subtracted spectra we determine the nature and value of the bandgap of polytypic WS₂, finding E_BG=1.6_-0.2^+0.3eV with a clear preference for an indirect bandgap. Further, we demonstrate how this method enables us to robustly identify excitonic transitions down to very small energy losses. Our approach has been implemented and made available in an open source Python package dubbed EELSfitter.

Deposition Kinetics and Compositional Control of Vacuum-Processed CH3NH3PbI3 Perovskite

Halide perovskites have generated considerable research interest due to their excellent optoelectronic properties in the past decade. To ensure the formation of high-quality semiconductors, the deposition process for the perovskite film is a critical issue. Vacuum-based processing is considered to be a promising method, allowing, in principle, for uniform deposition on a large area. One of the benefits of vacuum processing is the control over the film composition through the use of quartz crystal microbalances (QCMs) that monitor the rates of the components in situ. In metal halide perovskites, however, one frequently employed component or precursor, CH₃NH₃I, exhibits nonstandard sublimation properties. Here, we study in detail the sublimation properties of CH₃NH₃I and demonstrate that by correcting for its complex adsorption properties and by modeling the film growth, accurate predictions of the stoichiometry of the final perovskite film can be obtained.

Nanoscale light element identification using machine learning aided STEM-EDS

Light element identification is necessary in materials research to obtain detailed insight into various material properties. However, reported techniques, such as scanning transmission electron microscopy (STEM)-energy dispersive X-ray spectroscopy (EDS) have inadequate detection limits, which impairs identification. In this study, we achieved light element identification with nanoscale spatial resolution in a multi-component metal alloy through unsupervised machine learning algorithms of singular value decomposition (SVD) and independent component analysis (ICA). Improvement of the signal-to-noise ratio (SNR) in the STEM-EDS spectrum images was achieved by combining SVD and ICA, leading to the identification of a nanoscale N-depleted region that was not observed in as-measured STEM-EDS. Additionally, the formation of the nanoscale N-depleted region was validated using STEM–electron energy loss spectroscopy and multicomponent diffusional transformation simulation. The enhancement of SNR in STEM-EDS spectrum images by machine learning algorithms can provide an efficient, economical chemical analysis method to identify light elements at the nanoscale.

Towards the understanding of the gold interaction with AIII-BV semiconductors at the atomic level

AIII-BV semiconductors have been considered to be a promising material for decades in overcoming the limitations of silicon semiconductor devices. One of the important aspects within the AIII-BV semiconductor technology is gold-semiconductor interactions on the nanoscale. We report on the investigations into the basic chemical interactions of Au atoms with AIII-BV semiconductor crystals by the investigation of the nanostructure formation in the process of thermally-induced Au self-assembly on various AIII-BV surfaces by means of atomically resolved High Angle Annular Dark Field (HAADF) Scanning Transmission Electron Microscopy (STEM) measurements. We have found that the formation of nanostructures is a consequence of the surface diffusion and nucleation of adatoms produced by Au induced chemical reactions on AIII-BV semiconductor surfaces. Only for InSb crystals we have found that there is efficient diffusion of Au atoms into the bulk, which we experimentally studied by Machine Learning HAADF STEM image quantification and theoretically by Density Functional Theory (DFT) calculations with the inclusion of finite temperature effects. Furthermore, the effective number of Au atoms needed to release one AIII metal atom has been estimated. The experimental finding reveals a difference in the Au interactions with the In- and Ga-based groups of AIII-BV semiconductors. Our comprehensive and systematic studies uncover the details of the Au interactions with the AIII-BV surface at the atomic level with chemical sensitivity and shed new light on the fundamental Au/AIII-BV interactions at the atomic scale.

Machine Learning in Nanoscience: Big Data at Small Scales

Recent advances in machine learning (ML) offer new tools to extract new insights from large data sets and to acquire small data sets more effectively. Researchers in nanoscience are experimenting with these tools to tackle challenges in many fields. In addition to ML's advancement of nanoscience, nanoscience provides the foundation for neuromorphic computing hardware to expand the implementation of ML algorithms. In this Mini Review, we highlight some recent efforts to connect the ML and nanoscience communities by focusing on three types of interaction: (1) using ML to analyze and extract new insights from large nanoscience data sets, (2) applying ML to accelerate material discovery, including the use of active learning to guide experimental design, and (3) the nanoscience of memristive devices to realize hardware tailored for ML. We conclude with a discussion of challenges and opportunities for future interactions between nanoscience and ML researchers.

Multivariate Statistical Analysis on a SEM/EDS Phase Map of Rare Earth Minerals

The scanning electron microscope/X-ray energy dispersive spectrometer (SEM/EDS) system is widely applied to rare earth minerals (REMs) to qualitatively describe their mineralogy and quantitatively determine their composition. The performance of multivariate statistical analysis on the EDS raw dataset can enhance the efficiency and the accuracy of phase identification. In this work, the principal component analysis (PCA) and the blind source separation (BSS) algorithms were performed on an EDS map of a REM sample, assisting to achieve an efficient phase map analysis. The PCA significantly denoised the phase map and was used as a preprocessing step for the following BSS. The BSS separated the mixed EDS signals into a set of physically interpretable components, bringing convenience to the phase separation and identification. Through the comparison between the independent component analysis (ICA) and the nonnegative matrix factorization (NMF) algorithms, the NMF was confirmed to be more suitable for the EDS mapping analysis.

Automatic microscopic image analysis by moving window local Fourier Transform and Machine Learning

Analysis of microscope images is a tedious work which requires patience and time, usually done manually by the microscopist after data collection. The results obtained in such a way might be biased by the human who performed the analysis. Here we introduce an approach of automatic image analysis, which is based on locally applied Fourier Transform and Machine Learning methods. In this approach, a whole image is scanned by a local moving window with defined size and the 2D Fourier Transform is calculated for each window. Then, all the Local Fourier Transforms are fed into Machine Learning processing. Firstly, a number of components in the data is estimated from Principal Component Analysis (PCA) Scree Plot performed on the data. Secondly, the data are decomposed blindly by Non-Negative Matrix Factorization (NMF) into interpretable spatial maps (loadings) and corresponding Fourier Transforms (factors). As a result, the microscopic image is analyzed and the features on the image are automatically discovered, based on the local changes in Fourier Transform, without human bias. The user selects only a size and movement of the scanning local window which defines the final analysis resolution. This automatic approach was successfully applied to analysis of various microscopic images with and without local periodicity i.e. atomically resolved High Angle Annular Dark Field (HAADF) Scanning Transmission Electron Microscopy (STEM) image of Au nanoisland of fcc and Au hcp phases, Scanning Tunneling Microscopy (STM) image of Au-induced reconstruction on Ge(001) surface, Scanning Electron Microscopy (SEM) image of metallic nanoclusters grown on GaSb surface, and Fluorescence microscopy image of HeLa cell line of cervical cancer. The proposed approach could be used to automatically analyze the local structure of microscopic images within a time of about a minute for a single image on a modern desktop/notebook computer and it is freely available as a Python analysis notebook and Python program for batch processing.

Unmixing noisy co-registered spectrum images of multicomponent nanostructures

Analytical electron microscopy plays a key role in the development of novel nanomaterials. Electron energy-loss spectroscopy (EELS) and energy-dispersive X-ray spectroscopy (EDX) datasets are typically processed to isolate the background-subtracted elemental signal. Multivariate tools have emerged as powerful methods to blindly map the components, which addresses some of the shortcomings of the traditional methods. Here, we demonstrate the superior performance of a new multivariate optimization method using a challenging EELS and EDX dataset. The dataset was recorded from a spectrum image P-type metal-oxide-semiconductor stack with 7 components exhibiting heavy spectral overlap and a low signal-to-noise ratio. Compared to peak integration, independent component analysis, Baysian Linear Unmixing and Non-negative matrix factorization, the method proposed was the only one to identify the EELS spectra of all 7 components with the corresponding abundance profiles. Using the abundance of each component, it was possible to retrieve the EDX spectra of all the components, which were otherwise impossible to isolate, regardless of the method used. We expect that this robust method will bring a significant improvement for the chemical analysis of nanomaterials, especially for weak signals, dose-sensitive specimen or signals suffering heavy spectral overlap.

Spectrum adapted the expectation-maximization algorithm for high-throughput peak shift analysis

We introduce a spectrum-adapted expectation-maximization (EM) algorithm for high-throughput analysis of a large number of spectral datasets by considering the weight of the intensity corresponding to the measurement energy steps. Proposed method was applied to synthetic data in order to evaluate the performance of the analysis accuracy and calculation time. Moreover, the proposed method was performed to the spectral data collected from graphene and MoS₂ field-effect transistors devices. The calculation completed in less than 13.4 s per set and successfully detected systematic peak shifts of the C 1s in graphene and S 2p in MoS₂ peaks. This result suggests that the proposed method can support the investigation of peak shift with two advantages: (1) a large amount of data can be processed at high speed; and (2) stable and automatic calculation can be easily performed.

An artificial intelligence atomic force microscope enabled by machine learning

Artificial intelligence (AI) and machine learning have promised to revolutionize the way we live and work, and one of the particularly promising areas for AI is image analysis. Nevertheless, many current AI applications focus on the post-processing of data, while in both materials sciences and medicine, it is often critical to respond to the data acquired on the fly. Here we demonstrate an artificial intelligence atomic force microscope (AI-AFM) that is capable of not only pattern recognition and feature identification in ferroelectric materials and electrochemical systems, but can also respond to classification via adaptive experimentation with additional probing at critical domain walls and grain boundaries, all in real time on the fly without human interference. Key to our success is a highly efficient machine learning strategy based on a support vector machine (SVM) algorithm capable of high fidelity pixel-by-pixel recognition instead of relying on the data from full mapping, making real time classification and control possible during scanning, with which complex electromechanical couplings at the nanoscale in different material systems can be resolved by AI. For AFM experiments that are often tedious, elusive, and heavily rely on human insight for execution and analysis, this is a major disruption in methodology, and we believe that such a strategy empowered by machine learning is applicable to a wide range of instrumentations and broader physical machineries.