Recent Progresses in Machine Learning Assisted Raman Spectroscopy

Qi, Yaping; Hu, Dan; Jiang, Yucheng; Wu, Zhenping; Zheng, Ming; Chen, Esther Xinyi; Liang, Yong; Sadi, Mohammad A.; Zhang, Kang; Chen, Yong P.

doi:10.1002/adom.202203104

Cited by 61 publications

(40 citation statements)

References 201 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…ML algorithms are able to learn and make predictions without being explicitly programmed and have some advantages over traditional chemometric analyses, like the ability to efficiently analyze large data sets and identify complex patterns. 128 A recent study demonstrated the use of a deep-learning model, a type of ML where algorithms create a network to store data and to learn, to denoise mass data sets. Researchers have demonstrated both supervised and unsupervised deep-learning algorithms, whereby a supervised approach requires high-quality data for training, and unsupervised techniques can be trained with noisier data.…”

Section: Data Analysis Methodsmentioning

confidence: 99%

“…More recently, machine learning (ML) techniques have been adopted to aid in Raman data analysis. ML algorithms are able to learn and make predictions without being explicitly programmed and have some advantages over traditional chemometric analyses, like the ability to efficiently analyze large data sets and identify complex patterns . A recent study demonstrated the use of a deep-learning model, a type of ML where algorithms create a network to store data and to learn, to denoise mass data sets.…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Mastering Femtosecond Stimulated Raman Spectroscopy: A Practical Guide

Lynch,

Das,

Alam

et al. 2023

ACS Phys. Chem Au

View full text Add to dashboard Cite

Femtosecond stimulated Raman spectroscopy (FSRS) is a powerful nonlinear spectroscopic technique that probes changes in molecular and material structure with high temporal and spectral resolution. With proper spectral interpretation, this is equivalent to mapping out reactive pathways on highly anharmonic excited-state potential energy surfaces with femtosecond to picosecond time resolution. FSRS has been used to examine structural dynamics in a wide range of samples, including photoactive proteins, photovoltaic materials, plasmonic nanostructures, polymers, and a range of others, with experiments performed in multiple groups around the world. As the FSRS technique grows in popularity and is increasingly implemented in user facilities, there is a need for a widespread understanding of the methodology and best practices. In this review, we present a practical guide to FSRS, including discussions of instrumentation, as well as data acquisition and analysis. First, we describe common methods of generating the three pulses required for FSRS: the probe, Raman pump, and actinic pump, including a discussion of the parameters to consider when selecting a beam generation method. We then outline approaches for effective and efficient FSRS data acquisition. We discuss common data analysis techniques for FSRS, as well as more advanced analyses aimed at extracting small signals on a large background. We conclude with a discussion of some of the new directions for FSRS research, including spectromicroscopy. Overall, this review provides researchers with a practical handbook for FSRS as a technique with the aim of encouraging many scientists and engineers to use it in their research.

show abstract

Section: Data Analysis Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Mastering Femtosecond Stimulated Raman Spectroscopy: A Practical Guide

Lynch,

Das,

Alam

et al. 2023

ACS Phys. Chem Au

View full text Add to dashboard Cite

show abstract

“…Anomaly detection methods based on autoencoders and reconstruction error were employed to identify abnormal Raman spectra, leading to promising classification results. 319 ML algorithms can also be used to analyze the large NMR data sets generated by TOCSY (total correlation spectroscopy) for identifying spin−spin coupling interactions and providing a detailed map of proton−proton connectivities, streamlining data analysis, and accelerating structure elucidation. This provides the complete structure of each compound in the mixture without having to separate the components physically or do a multivariate analysis.…”

Section: Continuum-scale Chemical Mixturesmentioning

confidence: 99%

“…This approach enables cost-effective spectrum prediction for large data sets and facilitates spectral feature identification. Anomaly detection methods based on autoencoders and reconstruction error were employed to identify abnormal Raman spectra, leading to promising classification results . ML algorithms can also be used to analyze the large NMR data sets generated by TOCSY (total correlation spectroscopy) for identifying spin–spin coupling interactions and providing a detailed map of proton–proton connectivities, streamlining data analysis, and accelerating structure elucidation.…”

Section: Continuum-scale Chemical Mixturesmentioning

confidence: 99%

Mixtures Recomposition by Neural Nets: A Multidisciplinary Overview

Nicolle,

Deng,

Ihme

et al. 2024

J. Chem. Inf. Model.

View full text Add to dashboard Cite

Artificial Neural Networks (ANNs) are transforming how we understand chemical mixtures, providing an expressive view of the chemical space and multiscale processes. Their hybridization with physical knowledge can bridge the gap between predictivity and understanding of the underlying processes. This overview explores recent progress in ANNs, particularly their potential in the ’recomposition’ of chemical mixtures. Graph-based representations reveal patterns among mixture components, and deep learning models excel in capturing complexity and symmetries when compared to traditional Quantitative Structure–Property Relationship models. Key components, such as Hamiltonian networks and convolution operations, play a central role in representing multiscale mixtures. The integration of ANNs with Chemical Reaction Networks and Physics-Informed Neural Networks for inverse chemical kinetic problems is also examined. The combination of sensors with ANNs shows promise in optical and biomimetic applications. A common ground is identified in the context of statistical physics, where ANN-based methods iteratively adapt their models by blending their initial states with training data. The concept of mixture recomposition unveils a reciprocal inspiration between ANNs and reactive mixtures, highlighting learning behaviors influenced by the training environment.

show abstract

“…Furthermore, an additional incentive for implementing RS in operando experiments is represented by the growing use of machine learning techniques to decode Raman data and contribute to expanding the existing spectral databases [56,57].…”

Section: Introductionmentioning

confidence: 99%

In situ and operando Raman spectroscopy of semiconducting photoelectrodes and devices for photoelectrochemistry

Favaro,

Kong,

Gottesman

2023

J. Phys. D: Appl. Phys.

View full text Add to dashboard Cite

Future alternative and promising energy sources involve photoelectrochemical (PEC) devices that can convert sunlight and abundant resources such as water and CO2 into chemical fuels and value-added products. However, identifying suitable photoabsorber semiconductor materials that fulfill all the stringent requirements of photoelectrodes in PEC devices remains a significant challenge. A key factor for tailoring and optimizing existing and novel photoabsorbers is understanding the processes occurring at the semiconductor/liquid electrolyte interface under working conditions. This perspective focuses on the application of operando Raman spectroscopy (RS) in synergy with (photo)electrochemical techniques. Despite being a relatively new field of application, when applied to photoelectrochemistry, operando RS offers insights into the evolution of photoelectrode structure (i.e. phase purity and degree of crystallinity) and surface defects under working conditions. The challenges associated with operando RS for (photo)electrochemical applications, including the low quantum efficiency of inelastic scattering and fluorescence, and possible mitigation strategies are discussed. Furthermore, practical aspects such as sample/reactor geometry requirements and the surrounding environment of the photoelectrode sample during operando RS under PEC conditions are reviewed. We demonstrate that operando RS can be used to perform product analysis of solar-driven biomass reforming reactions, showing the approach’s limitations and discussing possible solutions to overcome them. This work concludes with a discussion on the current state of operando RS of semiconducting photoelectrodes and devices for photoelectrochemistry. We show a new methodology for performing operando RS with illumination resembling AM1.5 conditions and with time resolution spanning from tens to hundreds of milliseconds, suitable timescales for real-time monitoring of chemical reactions and degradation mechanisms occurring at the photoelectrode under investigation.

show abstract

Recent Progresses in Machine Learning Assisted Raman Spectroscopy

Cited by 61 publications

References 201 publications

Mastering Femtosecond Stimulated Raman Spectroscopy: A Practical Guide

Mastering Femtosecond Stimulated Raman Spectroscopy: A Practical Guide

Mixtures Recomposition by Neural Nets: A Multidisciplinary Overview

In situ and operando Raman spectroscopy of semiconducting photoelectrodes and devices for photoelectrochemistry

Contact Info

Product

Resources

About