The accelerating convergence of artificial intelligence, machine learning, and chemometrics has created both opportunity and confusion within the analytical sciences. While many AI concepts are mathematically and statistically aligned with classical multivariate methods, differences in terminology and framing often obscure their shared foundations. This glossary presents a more comprehensive, technically rigorous mapping between AI and chemometric concepts, emphasizing latent-variable modeling, probabilistic inference, calibration theory, optimization, and uncertainty estimation. Each term is defined in its native AI context and explicitly related to its chemometric analogue, highlighting continuity rather than disruption. By unifying vocabularies across disciplines, this glossary serves as a practical reference, an educational resource, and a conceptual bridge for scientists applying advanced data-driven models to spectroscopic and analytical measurements.
This article was presented in The James L. Waters Symposium: Generative AI in the Analytical Chemist’s Toolbox for Chemical Measurements. Analytical spectroscopy has historically relied on chemometric methods such as principal component analysis (PCA) and partial least squares (PLS) to extract quantitative chemical information from complex spectral measurements. Recent advances in artificial intelligence—particularly generative modeling—extend these foundations by learning the full statistical structure of chemical measurement data. Generative models such as variational autoencoders (VAEs), generative adversarial networks (GANs), diffusion models, and transformer architectures enable the simulation of spectra, modeling of measurement uncertainty, calibration transfer between instruments, and even the prediction of molecular structures from spectral fingerprints. This paper examines the conceptual continuity between classical chemometrics and modern generative AI, illustrating how latent-variable philosophy underlies both approaches. The emerging integration of chemometrics, machine learning, and generative AI is transforming spectroscopy from a purely predictive discipline toward one capable of simulation, interpretation, and discovery.
The use of insulin injections to control one’s blood glucose level provides life-saving control for patients with diabetes. In its native state, after proinsulin is converted to insulin by cleavage and formation of three disulfide bonds, the insulin is stored in the pancreas as a hexamer that is stabilized with two zinc ions. When needed (because the glucose level in blood rises) insulin is released from the pancreas into the blood stream and disassociates into monomers which are the biologically active forms. Pharmaceutical companies have engineered modified insulins in order to optimize its control time by making point changes to a few amino acids. However, maintenance of the insulin in its active form is of prime importance because modification of insulin (or almost any protein) can cause aggregation into non-active or toxic forms. For these reasons information on the physical-chemical form of the molecules is of prime importance to the pharmaceutical corporations who are manufacturing these products and selling them to patients. It is known that degraded insulin forms amyloid fibrils that are high in β sheet protein structure. This column will show Raman results and ATEEM fluorescence whose correlations indicate that important information is available non-destructively.
Artificial intelligence methods are increasingly applied to spectroscopic data analysis, often framed as disruptive alternatives to traditional chemometrics. This article presents a chemometric interpretation of AI, demonstrating that modern machine learning (ML) approaches largely generalize familiar multivariate statistical concepts rather than replace them. Supervised, unsupervised, generative, and multimodal AI methods are examined through the lens of regression, latent-variable modeling, preprocessing, and forward modeling. Emphasis is placed on interpretability, model diagnostics, and chemical meaning, illustrating how AI extends classical chemometrics to nonlinear, large-scale, and heterogeneous spectroscopic systems.
This year’s Emerging Leader in Atomic Spectroscopy Award recipient is Sarah Theiner, whose research is focused on the application of atomic spectroscopy techniques—laser ablation inductively coupled plasma–mass spectrometry (LA-ICP-MS) and single-cell ICP-MS—to expand these analytical techniques as tools for biological and clinical imaging and drug-distribution studies.
The convergence of spectroscopy and generative artificial intelligence represents one of the most significant paradigm shifts in analytical science in the past decade. In 2025, several influential publications advanced this convergence from exploratory demonstrations to broadly applicable methodologies. This article critically reviews six influential works that help define the landscape of generative artificial intelligence in spectroscopy, spanning synthetic spectral generation, data augmentation, inverse molecular inference, physics-informed modeling, and transformer-based structural elucidation. Together, these studies inaugurate generative artificial intelligence as a foundational tool for next-generation spectroscopic analysis, calibration, and discovery.
Now that our Grand Review is over it is time to move on to what I call “advanced topics.” This will be a series of articles diving into more detail on functional groups previously studied and will introduce new functional groups not yet discussed. In this column we will briefly review the C-H stretching and bending vibrations of methyl and methylene groups and then discuss how to use infrared spectroscopy to determine some of the branch points found in alkanes. We will see that the key is that some branch points give rise to what is known as a “split umbrella mode.”
Spectroscopy instrumentation and software are transitioning to intelligent, interconnected analytical "ecosystems." Advances in detection, optics, and software across electronic, vibrational, atomic, and imaging spectroscopy have resulted in higher sensitivity, miniaturized and portable platforms, and multimodal capabilities. New technologies such as quantum-enhanced detectors, ultra-fast electronics, high-power lasers, and artificial intelligence (AI)-driven predictive analytics have transformed instrument performance. Simultaneously, chemometric and AI-driven tools are enabling predictive modeling, automated classification, and real-time process analytics. This review synthesizes major developments in spectroscopy systems and software from 2025-2026 and discusses emerging trends toward autonomous analytical laboratories.
In our ongoing review of infrared spectra, we will study organic nitrogen containing compounds including amides and amines. Amides contain both nitrogen and a C=O group and are found in proteins and polymers. Amines contain carbon, nitrogen, and hydrogen, and are ubiquitous in medicines. As always, concepts will be illustrated with reference spectra.
The primary goal of this study was to evaluate two microwave digestion systems for the acid decomposition of biological tissues: (1) a single reaction chamber (SRC) system and (2) a rotor-based, closed vessel microwave (RBCVM) system. Four key toxic trace elements, arsenic (As), cadmium (Cd), lead (Pb) and mercury (Hg), were determined in seven certified reference materials (CRMs), and one other RM by Inductively Coupled Plasma Tandem Mass spectrometry (ICP-MS/MS). Results for both digestion systems were within +/- 10% for As, Cd, and Pb, and within +/- 20% for Hg compared to certified or reference values and were deemed "fit-for-purpose." Measurement repeatability was largely comparable between the two systems. The conclusion of the study was that the performance of the two digestion systems was deemed equivalent for As, Cd, Pb, and Hg in the eight materials evaluated.
Near-infrared (NIR) spectroscopy combined with aquaphotomics shows potential for a rapid, non-invasive approach for detecting subtle biochemical changes in biofluids and agricultural products. By monitoring water molecular structures through water matrix coordinates (WAMACs) and visualizing water absorption spectrum patterns (WASPs) via aquagrams, researchers can identify disease biomarkers, food contaminants, and other analytes with high accuracy. This tutorial introduces the principles, practical workflow, and applications of NIR aquaphotomics for everyday laboratory use.
Artificial intelligence (AI) is promising to redefine vibrational spectroscopy in 2025, marking a major inflection point in spectroscopic analysis, calibration, and interpretation across Raman, infrared (IR), near-infrared (NIR), and hyperspectral imaging platforms. From agricultural sensing to precision oncology, the fusion of machine learning (ML), deep neural networks (NNs), quantile regression forests, and explainable AI (XAI) is transforming spectroscopic workflows into autonomous, scalable, and predictive modeling systems (1–3). New software platforms such as SpectrumLab and SpectraML demonstrate how generative models, foundation architectures, and physics-informed neural networks can automate feature extraction and deliver predictive models with actionable uncertainty estimates (1–2,4). The following review consolidates trends across 30 Spectroscopy publications from 2025, offering a narrative on how AI is enabling innovation in data fusion, spectral imaging, industrial bioprocess monitoring, precision agriculture, environmental risk assessment, biomedical diagnostics, and other applications.
Fourier transform infrared (FT-IR) spectroscopy is a versatile, non-destructive analytical tool used to characterize molecular structures, monitor chemical reactions, and quantify analytes in diverse materials. This mini-tutorial reviews fundamental principles, key operational modes, and practical examples across environmental, biomedical, and industrial applications. Readers will review and learn how to optimize FT-IR methods, interpret spectra, and avoid common pitfalls in data collection and processing.
In 2025, the vibrational-spectroscopy community saw a convergence of deep learning, advanced simulation, and portable instrumentation that materially changed how spectra are interpreted and applied. Breakthroughs in spectrum-to-structure models, machine learning (ML)-accelerated molecular dynamics, and field-deployable classic Raman, near-infrared (NIR), and surface-enhanced Raman spectroscopy (SERS) sensors pushed vibrational techniques from complex laboratory characterization toward automated structure elucidation, rapid analysis, and real-world sample sensing. This summary article highlights key 2025 contributions and their implications for the year of discovery.
This paper aims to identify gemstones using various spectroscopic methods, including Raman spectroscopy and spectrophotometry, as well as spectral imaging techniques like hyperspectral and multi-band imaging.
In this part of our ongoing review of the infrared spectra of carbonyl-containing functional groups, we will study the spectra of esters and carbonates. Esters are ubiquitous in our food and medicines, and polymeric carbonates form an important part of the materials around us. As always, concepts will be illustrated with reference spectra.
Sports drinks, also known as electrolyte drinks, are a popular beverage choice among consumers and come in a wide variety of flavors and electrolyte compositions. They may vary substantially in their additives, which typically include sweeteners, coloring agents, flavoring agents, and additional vitamins and nutrients. The determination of electrolyte elements is important for accurate product labeling and quality control; however, conventional instrumentation, such as inductively coupled plasma optical emission spectroscopy (ICP-OES), may be costly to both acquire and operate, especially given its relatively high consumption of argon gas. Here we present a rapid and cost-effective method using microwave plasma atomic emission spectroscopy (MP-AES), which utilizes a nitrogen plasma that may be supplied with either a conventional gas source (dewar or cylinder gas) or a nitrogen generator on-site. Using a simple “dilute and shoot” method, electrolyte elements can be determined without prior sample digestion with good reproducibility and excellent limits of detection across a variety of sample matrices. Thus, MP-AES offers a simple multi-element alternative to ICP-OES for sports drink analysis without costly argon consumption.
This month, I am going to talk about a different spectroscopy that rivals Raman in its usefulness. A-TEEM stands for Absorption, Transmission, Excitation/Emission Matrix Fluorescence. While it does not give direct information on the structure of a molecule, its value lies in its ability to detect subtle changes in structure due to changes in conformation of large organic molecules or changes in interactions with other molecules. While this is not a technology with which I have been directly involved, I have been impressed with its potential and thought this column would be a good place to share my thoughts.
Phytotoxins, such as aconitine, have a dual function of bioactivity and toxicity, which intentionally or unintentionally trigger the food poisoning associated with them. The detection of toxins and poisons in complex matrices is challenging, and rapid accurate characterization is a prerequisite for tracing the source of poisoning and choosing the correct treatment. In this study, surface-enhanced Raman spectroscopy (SERS) and solvent extraction were used to detect aconitine (AC) in various complex matrices using gold nanorod substrates. The experimental results demonstrated that ether efficiently extracted AC from soy sauce as an example complex matrix. When performed using a handheld Raman spectrometer, the entire detection process was completed in 5s, and the detection limit was 1 ppm. This rapid, simple, and effective method enables sensitive detection of toxicants in various complex matrices and has potential applications in emergency response and public safety.
It is now well-known that Raman spectroscopy is being used for tracking the structural form of carbons used in industrial applications. The structure of graphite, microcrystalline graphite, and the nanotubes and buckeyballs are well understood and described by rigorous physical methods. However, many of the carbon materials used in large commercial applications are not as well characterized. Depending on the substrates from which they are made, they can be graphitic, graphitizable, or non-graphitizable. For instance, the preferred carbon for use in the anodes of Li+ batteries is termed hard carbon, and its properties depend on the selection of starting materials and processing conditions. In this column, we will describe what is known about the structures of these materials and how Raman spectroscopy can characterize them. If you follow my columns, you will realize that this is not the first column that I have written on the Raman spectra of carbons. In this case, the ultimate goal is to provide our readers enough information for them develop correlations between the Raman spectral characteristics and performance of the materials.
