RESUMEN
In recent years machine learning has made extensive progress in modeling many aspects of mass spectrometry data. We brought together proteomics data generators, repository managers, and machine learning experts in a workshop with the goals to evaluate and explore machine learning applications for realistic modeling of data from multidimensional mass spectrometry-based proteomics analysis of any sample or organism. Following this sample-to-data roadmap helped identify knowledge gaps and define needs. Being able to generate bespoke and realistic synthetic data has legitimate and important uses in system suitability, method development, and algorithm benchmarking, while also posing critical ethical questions. The interdisciplinary nature of the workshop informed discussions of what is currently possible and future opportunities and challenges. In the following perspective we summarize these discussions in the hope of conveying our excitement about the potential of machine learning in proteomics and to inspire future research.
Asunto(s)
Aprendizaje Automático , Proteómica , Proteómica/métodos , Algoritmos , Espectrometría de MasasRESUMEN
Controlling materials' morphology, crystal phase and chemical composition at the atomic scale has become central in materials research. Wet chemistry approaches have great potential in directing the material crystallisation process to achieve tuneable chemical compositions as well as to target specific crystal phases. Herein, we report the compositional and crystal phase tuneability achieved in the quasi-binary WxMo1-xSe2 system with chemical and crystal phase mixing down to the atomic level. A series of WxMo1-xSe2 solid solutions in the form of nanoflowers with atomically thin petals were obtained via a direct colloidal reaction by systematically varying the ratios of transition metal precursors. We investigate the effect of selenium precursor on the morphology of the WxMo1-xSe2 material and show how using elemental selenium can enable the formation of larger and distinct nanoflowers. While the synthesised materials are compositionally homogeneous, they exhibit crystal phase heterogeneity with the co-existing domains of the 1T' and 2H crystal phases, and with evidence of MoSe2 in the metastable 1T' phase. We show at single atom level of resolution, that tungsten and molybdenum can be found in both the 1T' and 2H lattices. The formation of heterophase 1T'/2H WxMo1-xSe2 electrocatalysts allowed for a considerable improvement in the activity for the acidic hydrogen evolution reaction (HER) compared to pristine, 1T'-dominated, WSe2. This work can pave the way towards engineered functional nanomaterials where properties, such as electronic and catalytic, have to be controlled at the atomic scale.
RESUMEN
BACKGROUND: Machine learning (ML) technologies, especially deep learning (DL), have gained increasing attention in predictive mass spectrometry (MS) for enhancing the data-processing pipeline from raw data analysis to end-user predictions and rescoring. ML models need large-scale datasets for training and repurposing, which can be obtained from a range of public data repositories. However, applying ML to public MS datasets on larger scales is challenging, as they vary widely in terms of data acquisition methods, biological systems, and experimental designs. RESULTS: We aim to facilitate ML efforts in MS data by conducting a systematic analysis of the potential sources of variability in public MS repositories. We also examine how these factors affect ML performance and perform a comprehensive transfer learning to evaluate the benefits of current best practice methods in the field for transfer learning. CONCLUSIONS: Our findings show significantly higher levels of homogeneity within a project than between projects, which indicates that it is important to construct datasets most closely resembling future test cases, as transferability is severely limited for unseen datasets. We also found that transfer learning, although it did increase model performance, did not increase model performance compared to a non-pretrained model.
Asunto(s)
Aprendizaje Automático , Espectrometría de Masas en Tándem , Cromatografía LiquidaRESUMEN
Tuning the charge transport properties of two-dimensional transition metal dichalcogenides (TMDs) is pivotal to their future device integration in post-silicon technologies. To date, co-doping of TMDs during growth still proves to be challenging, and the synthesis of doped WSe2, an otherwise ambipolar material, has been mainly limited to p-doping. Here, we demonstrate the synthesis of high-quality n-type monolayered WSe2 flakes using a solid-state precursor for Se, zinc selenide. n-Type transport has been reported with prime electron mobilities of up to 10 cm2 V-1 s-1. We also demonstrate the tuneability of doping to p-type transport with hole mobilities of 50 cm2 V-1 s-1 after annealing in air. n-Doping has been attributed to the presence of Zn adatoms on the WSe2 flakes as revealed by X-ray photoelectron spectroscopy (XPS), spatially resolved time of flight secondary ion mass spectroscopy (SIMS) and angular dark-field scanning transmission electron microscopy (AD-STEM) characterization of WSe2 flakes. Monolayer WSe2 flakes exhibit a sharp photoluminescence (PL) peak at room temperature and highly uniform emission across the entire flake area, indicating a high degree of crystallinity of the material. This work provides new insight into the synthesis of TMDs with charge carrier control, to pave the way towards post-silicon electronics.
RESUMEN
Crystal phase control in layered transition metal dichalcogenides is central for exploiting their different electronic properties. Access to metastable crystal phases is limited as their direct synthesis is challenging, restricting the spectrum of reachable materials. Here, we demonstrate the solution phase synthesis of the metastable distorted octahedrally coordinated structure (1T' phase) of WSe2 nanosheets. We design a kinetically-controlled regime of colloidal synthesis to enable the formation of the metastable phase. 1T' WSe2 branched few-layered nanosheets are produced in high yield and in a reproducible and controlled manner. The 1T' phase is fully convertible into the semiconducting 2H phase upon thermal annealing at 400 °C. The 1T' WSe2 nanosheets demonstrate a metallic nature exhibited by an enhanced electrocatalytic activity for hydrogen evolution reaction as compared to the 2H WSe2 nanosheets and comparable to other 1T' phases. This synthesis design can potentially be extended to different materials providing direct access of metastable phases.
RESUMEN
We report on spin transport in WS2-based 2D-magnetic tunnel junctions (2D-MTJs), unveiling a band structure spin filtering effect specific to the transition metal dichalcogenides (TMDCs) family. WS2 mono-, bi-, and trilayers are derived by a chemical vapor deposition process and further characterized by Raman spectroscopy, atomic force microscopy (AFM), and photoluminescence spectroscopy. The WS2 layers are then integrated in complete Co/Al2O3/WS2/Co MTJ hybrid spin-valve structures. We make use of a tunnel Co/Al2O3 spin analyzer to probe the extracted spin-polarized current from the WS2/Co interface and its evolution as a function of WS2 layer thicknesses. For monolayer WS2, our technological approach enables the extraction of the largest spin signal reported for a TMDC-based spin valve, corresponding to a spin polarization of PCo/WS2 = 12%. Interestingly, for bi- and trilayer WS2, the spin signal is reversed, which indicates a switch in the mechanism of interfacial spin extraction. With the support of ab initio calculations, we propose a model to address the experimentally measured inversion of the spin polarization based on the change in the WS2 band structure while going from monolayer (direct bandgap) to bilayer (indirect bandgap). These experiments illustrate the rich potential of the families of semiconducting 2D materials for the control of spin currents in 2D-MTJs.
RESUMEN
Monolayer TiS2 is the lightest member of the transition metal dichalcogenide family with promising applications in energy storage and conversion systems. The use of TiS2 has been limited by the lack of rapid characterization of layer numbers via Raman spectroscopy and its easy oxidation in wet environment. Here, we demonstrate the layer-number-dependent Raman modes for TiS2. 1T TiS2 presents two characteristics of the Raman active modes, A1g (out-of-plane) and Eg (in-plane). We identified a characteristic peak frequency shift of the Eg mode with the layer number and an unexplored Raman mode at 372 cm-1 whose intensity changes relative to the A1g mode with the thickness of the TiS2 sheets. These two characteristic features of Raman spectra allow the determination of layer numbers between 1 and 5 in exfoliated TiS2. Further, we develop a method to produce oxidation-resistant inks of micron-sized mono- and few-layered TiS2 nanosheets at concentrations up to 1 mg/mL. These TiS2 inks can be deposited to form thin films with controllable thickness and nanosheet density over square centimeter areas. This opens up pathways for a wider utilization of exfoliated TiS2 toward a range of applications.
RESUMEN
The rise of atomically thin materials has the potential to enable a paradigm shift in modern technologies by introducing multi-functional materials in the semiconductor industry. To date the growth of high quality atomically thin semiconductors (e.g. WS2) is one of the most pressing challenges to unleash the potential of these materials and the growth of mono- or bi-layers with high crystal quality is yet to see its full realization. Here, we show that the novel use of molecular precursors in the controlled synthesis of mono- and bi-layer WS2 leads to superior material quality compared to the widely used direct sulfidization of WO3-based precursors. Record high room temperature charge carrier mobility up to 52 cm2/Vs and ultra-sharp photoluminescence linewidth of just 36 meV over submillimeter areas demonstrate that the quality of this material supersedes also that of naturally occurring materials. By exploiting surface diffusion kinetics of W and S species adsorbed onto a substrate, a deterministic layer thickness control has also been achieved promoting the design of scalable synthesis routes.