This work utilized terahertz time-domain spectroscopy (THz-TDS) and Fourier Transform Infrared (FTIR) spectroscopy to characterize epoxy blends. The epoxy blends were prepared by blending a virgin DER-332/LC-100 epoxy system with Recovered Epoxy Polymer (REP) at 1-10 wt%, and were compared with references of virgin epoxy and REP. The REP was obtained via peracetic acid digestion of Carbon Fibre Reinforced Polymer (CFRP), followed by purification. Using THz-TDS amplitude-phase analysis, five optical parameters were obtained: refractive index (n), extinction coefficient (κ), absorption coefficient (α), and the real and imaginary of the dielectric constant (ε' and ε″). Under ambient conditions, these parameters extracted from THz-TDS were unable to distinguish low REP content 0-10% REP but were able to clearly differentiate 100% REP from all other epoxy blends. Compared to the other blends, 100% REP shows markedly increased n, κ, α, ε', and ε″, attributed to oxidation, formation of polar groups, chain scission and microvoid-induced scattering. To distinguish low REP-content blends using THz-TDS, external perturbation was applied. Under perturbation, linear concentration-dependent trends in Δn, Δκ, Δα, Δε' and Δε'' were observed. The external perturbation enhanced the interaction between the samples and the THz wave, likely due to amplification of dipole relaxation. Additionally, the transmission spectra (3500-7500 cm[-][1]) obtained from FTIR increased as the REP content increased from 0-10%. This concentration dependent trend can be attributed to partial network dilution and reduced overtone absorption. However, the transmission of 100% REP showed strong attenuation due to structural disorder and enhanced scattering.

As a traditional Chinese medicinal material, the quality of Poria cocos (PC) significantly affects its pharmacological efficacy. However, adulteration of PC in the market has been frequently observed. Adulterated PC not only compromises its medicinal value but also poses potential risks to public health. Conventional identification methods are often labor-intensive, destructive, and insufficiently sensitive to complex adulteration scenarios. In this study, a rapid and nondestructive strategy for both qualitative and quantitative analysis of PC adulteration is developed by integrating near-infrared (NIR) two-dimensional correlation spectroscopy (2DCOS) with deep learning algorithms. NIR spectra in the range of 960-1600 nm are collected from pure PC and samples adulterated with 5 common adulterants at concentration gradients of 5%. Specifically, to enhance spectral resolution, a systematic 2DCOS dataset is established by combining 5 representative spectral preprocessing methods with 2 types of 2DCOS. Furthermore, a bidirectional long short-term memory network with an attention mechanism (BiLSTM-Attn) is proposed to fully exploit the sequential dependency of correlation features embedded in 2DCOS maps. Compared with experiments involving convolutional neural network (CNN), gated recurrent unit (GRU), long short-term memory (LSTM), and bidirectional LSTM (BiLSTM) models, the proposed BiLSTM-Attn achieved 100% classification accuracy and superior regression performance, with a mean coefficient of determination (R[2]) of 0.9920, a low root mean squared error (RMSE) of 2.41, and a high mean residual predictive deviation (RPD) of 12.85. Extended experiments demonstrate that the combination of NIR 2DCOS and deep learning algorithms enables accurate and robust detection of PC adulteration. The proposed framework offers valuable support for quality control in traditional Chinese medicine, and has a wide range of applications.

A highly sensitive method has been developed for the determination of bisphenol A (BPA) in beverages (juice, beer, carbonated soft drinks) using surface-enhanced Raman spectroscopy (SERS) coupled with solid-phase extraction (SPE), in situ synthesis of silver nanoparticles (AgNPs), and surface modification with tetraoctylammonium bromide (TNOAB). SPE showed high BPA recovery rates (>90%) from complex food matrices. The crucial factor for achieving maximum sensitivity was the in situ synthesis of AgNPs directly in the presence of the analyte, as the use of pre-synthesized nanoparticles led to an approximately 100-fold decrease in signal intensity. Investigation of cationic modifiers with different structures revealed that the efficiency of SERS signal enhancement correlates with alkyl chain length. TNOAB (C8) provided a 7-fold increase in signal compared to the unmodified system, significantly outperforming the effects of tetrabutylammonium bromide (C4, 1.7-fold enhancement) and triethylamine (C2, 1.3-fold enhancement). The limit of detection for BPA was 0.077 ppb with a signal enhancement factor of 1.87×10[8]. For quantitative analysis, a PLS calibration model was applied in the range of 1-100 ppb (R[2] = 0.9986, RMSEC = 1.75 ppb, RMSEP = 1.95 ppb). The simplicity of the developed approach makes it suitable for routine monitoring of BPA content in beverages.

Near-Infrared Spectroscopy (NIRS) can directly provide product quality information at the molecular level, such as concentration, viscosity, etc. However, NIRS is overly sensitive to environmental noise. Therefore, how to make full use of the information of process variables to enhance the robustness of NIR modeling is the key. For this purpose, this paper presents a multi-level fusion model based on Attention Mechanism (AM). First, two-branch Convolutional Neural Network (CNN) is introduced to extract the feature of NIRS. Then, a fusion block based on AM is designed, which consists a cross-attention operation, a residual connection, and a Multi-layer Perceptron (MLP). The residual connection weights and integrates features of NIRS and process variables according to the cross-attention distribution, while the MLP further combines the features in the feature dimension. Additionally, considering different information exchanged ways of fused features at each level of two-stream CNN, two different fusion schemes are designed namely central and shared strategies. Experiments on the 2,6-dimethylphenol distillation process datasets demonstrate that proposed method outperforms some existing methods.

The X-ray fluorescence (XRF) spectroscopy is widely used in mining for elemental analysis, but it cannot detect lithium (Li) due to its low atomic number. The low sensitivity and high detection limits of conventional laser-induced breakdown spectroscopy (LIBS) also present serious challenges for analyzing lithium ore. This work demonstrates improved analytical capabilities of LIBS by using an external electric field to detect Li and other trace elements in Li ore. A notable increase in emission intensity and detection sensitivity for all trace elements has been observed in electric field-assisted LIBS (EF-LIBS) compared to conventional LIBS. Significant changes in plasma parameters, such as stabilized electron density and an extended plasma lifespan (∼10 μs), were achieved by applying the electric field, resulting in better spectral signals. The most impressive improvement was an eightfold increase in the detection limit for lithium, which decreased from 46.07 ppm (traditional LIBS) to 5.93 ppm (EF-LIBS). Increased excitation efficiency under the electric field and a longer plasma plume lifetime are responsible for this enhancement. For lithium exploration and mining, EF-LIBS offers a better alternative to traditional methods due to its field-deployable design and low sample requirements, providing increased sensitivity and operational advantages without the need for additional complex equipment.

The combination of laser-induced breakdown spectroscopy (LIBS) and machine learning provides a rapid approach for coal ash quantification. However, in practical coal preparation plants, calibration samples are often limited, and ash contents are usually confined to a narrow range, making reliable quantitative modeling difficult. To address this issue, we propose a structured spectral in-context learning framework for small-sample coal ash quantification. Unlike conventional methods that rely on task-specific iterative parameter optimization, the proposed framework performs downstream inference by conditioning on a limited set of calibration samples using a pre-trained tabular foundation model. To adapt continuous high-dimensional LIBS spectra to this framework, segmented statistical descriptors and multi-scale Haar-like features were constructed to capture subtle local spectral variations associated with narrow ash-content changes. The experimental results show that the proposed SS-ICL framework achieves robust predictive performance under restricted conditions, with an R[2] of 0.8880 and an RMSE of 0.1953 in a narrow-range washed-coal scenario. These results demonstrate that the proposed framework provides a practical and robust strategy for on-site LIBS-based coal quality analysis in restricted industrial environments.

Layered mineral composites may exhibit incidence-side-dependent terahertz responses because interfacial reflection and unequal optical paths modify wave propagation through the structure. In this study, terahertz time-domain spectroscopy (THz-TDS) was employed to investigate the effective refractive index response of opal-chlorite bilayer composites. The results show that the refractive index difference between opposite incidence sides decreases and gradually approaches a stable level with increasing chlorite thickness. TMM inversion retrieved the chlorite refractive index spectrum in good agreement with the directly measured result. Proton irradiation applied from the opal side enhanced the asymmetry mainly in the thinnest sample, whereas this effect was rapidly suppressed as the chlorite thickness increased. In addition, the measurements under the investigated environmental conditions showed that the refractive index difference remained comparatively stable. These results provide an experimental basis for analyzing incidence-side-dependent terahertz responses in layered mineral bilayers.

A divide-and-conquer approach is presented that helps undergraduate students interpret organic infrared-spectra. In this method, DivnConqIR, students deal with the confusion of a complicated spectrum by focusing first on the carbonyl band at ∼1720 cm[-1] near the middle of the spectrum. When present, its position and related bands can be analyzed to determine the specific functional group to which the carbonyl belongs; in its absence, students eliminate many functional groups. In either case, students begin spectral interpretation with significant accomplishment. This strategy divides the spectrum again, now at 3000 cm[-1], to define the backbone of the organic molecule and determine the possible presence of alkenes, alcohols, amines, or terminal alkynes. Students divide the spectrum again, now at 1375 cm[-1], to confirm the presence of methyl group(s) with the methyl umbrella band or, if present, isopropyl or tert-butyl doublet bands for further connectivity. Students divide the spectrum again, now at 1000 cm[-1], to search for the presence of one or more C-O stretches and to confirm the presence of alcohols, carboxylic acids, esters, or anhydrides, or to discover ethers. Lastly, students use DivnConqIR to examine unaccounted bands to identify other useful functional groups that may be present. Students interpret IR spectra without confusion and determine functional groups effectively; they integrate functional groups into possible complete structures. Qualitative and quantitative student performance data are offered to suggest the effectiveness of this interpretive approach, and DivnConqIR is compared to other IR interpretation methods to show that it is unique and most comprehensive. Many worked examples are presented, and additional resources are provided.

This review is concerned with our recent four papers on vibrational spectroscopy studies of water. The four papers had dual purposes. One was to deepen the understanding of IR, Raman and NIR spectra of water including the low-frequency regions of IR (far-infrared; FIR) and Raman spectra, and another was to explore the structure of water and water hydrogen-bonding networks. In this review we put emphasis on the former. Particularly, the importance of 2D-COS in the spectral analysis was explained in detail. In the studies of low-frequency region of IR spectra, 2D-COS revealed that there are two or three bands assignable to libration L2 modes of tetrahedron structure (THS) and pseudo-tetrahedron structure (PTHS). Bast on these observations we concluded that there are two or three water hydrogen bonding networks with THS or PTHS species. In the NIR study of water 2D-COS was used to construct the relations between PTHS and THS bands in the NIR spectra an those in the IR spectra. 2D-COS was also useful to determine the sequence order of the events occurring during temperature increase.

Bloodstains are among the most common forms of evidence at crime scenes and are considered one of the most critical types of physical evidence in forensic science. Accurately determining the age of bloodstains at a crime scene is crucial to solving cases. Traditionally, this determination relied on the expertise of forensic investigators or biochemical analysis methods, both of which often damaged the samples and increased investigation costs. This necessitated the development of a rapid, non-destructive, and accurate method for bloodstain age estimation. In the proposed method, bloodstains are applied to various substrates to minimize the influence of background interference on the experimental results. Spectral data are extracted non-destructively from the bloodstains and integrated with machine learning techniques to develop predictive models.The training model of blood aging was built using the Kernel Extreme Learning Machine (KELM) algorithm and the near-infrared (NIR) spectral data. It was also compared with the traditional Partial Least Squares Regression (PLS-R), Random Forest (RF) models and Back-propagation neural network (BP) models. The experimental results showed that the performance of KELM model was the best, test set with a determination coefficient (R[2]) of 0.94, which is higher compared to that of the PLS-R model (R[2] = 0.75), RF model (R[2] = 0.82) and BP model (R[2] = 0.90). Besides, the root mean square error of prediction (RMSEP) of the KELM model was 2.54 days, smaller than the PLS-R model (RMSEP = 5.18 days) , the RF model (RMSEP = 4.29days) and the BP model (RMSEP = 3.68 days).The mean absolute percentage error (MAPE) of the validation set was 2.45% when using the KELM algorithm, which was also the minimum of the three algorithms, and it represented that the prediction results were the best. The experimental data showed that the method proposed in this study could accurately estimate the age of bloodstains, providing a new technology reference for bloodstain detection.

Laser-induced Fluorescence (LIF) is widely employed in combustion research for various diagnostic applications, including the simultaneous visualisation of liquid and vapour phase as well as temperature and concentration measurements. For quantitative LIF diagnostics, liquid fuels are often doped with fluorescent tracers whose spectroscopic properties must be carefully characterised. While numerous dyes have been investigated in polar solvents like water or ethanol, data for nonpolar fluids like common combustion fuels remain scarce. However, solvent polarity is known to significantly affect solubility, their optical response, spectral shape as well as influencing shifts in absorption and emission maxima. This study provides a systematic comparison of the fluorescent dyes Coumarin 152, Fluoranthene, DCM, and Pyrromethene 567 with respect to their suitability quantitative two-colour LIF thermometry in combustion diagnostics and single-droplet evaporation experiments. Measurements are conducted in three nonpolar fuels: a fossil (B0) diesel and two synthetic alternatives, gas-to-liquid (GtL) and power-to-liquid (PtL) diesel. A systematic comparison of dye performance is presented, including solubility limits as well as temperature-dependent LIF response under 355 and 532 nm excitation. Despite its comparatively lower solubility, Pyrromethene 567 exhibits high signal intensities and a pronounced temperature-dependent response, qualifying it as suitable tracer for 2-colour LIF (2c-LIF) measurements in these fuels. The results support an informed selection of tracer-fuel combinations and contribute to improved reliability of LIF-based temperature and concentration measurements in nonpolar mixtures and real fuel application.

Efficient monitoring of solvent extraction processes is essential for the safe and sustainable operation of nuclear fuel reprocessing, particularly under conditions where radiolysis and solvent degradation can affect separation efficiency and phase stability. We investigate an integrated optical approach that combines Raman spectroscopy of the organic phase with laser-induced breakdown spectroscopy (LIBS) of the aqueous phase for simultaneous control of extractant composition and raffinate elemental content. Raman spectroscopy was applied to tributyl phosphate (TBP) in a hydrocarbon diluent over a range of concentrations selected to mimic changes in extractant loading and degradation. Using closely spaced bands in the 1000-1200 cm[-1] region, a simple ratiometric indicator yielded an excellent linear correlation with TBP concentration (R[2] = 0.996) and a median prediction error of about 1.0%. LIBS measurements were performed on aqueous solutions containing Zr, La, Ce and Sr as fission-product surrogates in concentration ranges representative of late extraction stages (1-40 g L[-1]). Despite shot-to-shot fluctuations associated with breakdown in liquids, internal standardization to solvent oxygen lines or a major solute element provided linear calibration curves with mean absolute percentage errors of 4-8% and limits of detection on the order of a few grams per liter. The combined Raman-LIBS scheme thus offers a promising basis for a reagent-free in situ monitoring of solvent extraction operations in nuclear fuel reprocessing.

Foodborne illnesses pose a serious threat to food safety and cause substantial economic losses. Hyperspectral imaging (HSI) has emerged as a powerful tool for rapid and non-destructive identification of foodborne pathogens. However, the high chemical similarity among different pathogen categories presents a challenge for accurate discrimination. To address this issue, we developed an optimized machine learning framework integrated with HSI that incorporates multimodal learning and a multi-head attention mechanism, enabling deeper extraction and fusion of spectral profiles and intensity images features. In visualization analyses, the deeply fused features demonstrated excellent inter-species separability, and the proposed multimodal multi-head attention fusion (MMAF) strategy achieved a high identification accuracy of 96.29%, representing an improvement of 4.17% over the single-modal approach. These results indicate that the optimized HSI approach, driven by advanced machine learning, holds great potential as an effective tool for rapid detection of microbial contamination in food products.

Mid-infrared spectroscopy is a powerful technique for probing molecular structure and composition, yet its application to aqueous systems remains limited by strong solvent absorption and the dynamic range of mid-IR detectors, which restrict the usable optical pathlength. Here, we present a quantum cascade laser-based spectroscopic setup that enables simultaneous acquisition of absorption spectra measured in transmission across a continuum of optical pathlengths using a custom-made wedge-shaped flow cell and a pyroelectric detector array. This design eliminates the need for mechanical adjustments or sample dilution, allowing direct analysis of complex liquid samples. We introduce the pathlength-to-noise ratio as a novel metric for determination of the optimal pathlength region for every wavenumber. By weighing detector signals based on this ratio, we achieve robust integration of multi-pathlength data into a single coherent mid-IR absorption spectrum. The system was validated using aqueous protein solutions of bovine serum albumin and β-lactoglobulin in the range of 1 to 25 mg mL[-1]. Characteristic amide I and II bands were successfully resolved, and secondary structure differences were detected. This work establishes a flexible approach to mid-IR absorption spectroscopy, aiming to eliminate the limitations of detectors with a small dynamic range and to get a better understanding of the optimal pathlength for complex samples.

Advanced ALD (atomic layer deposition) precursors are designed to deposit thin films that must be reliable, clean and free of contaminants. This created a need for analytical methods to measure contaminants and ensure the purity of the starting compounds used in ALD. However, accurate analysis of multiple trace impurity elements in high purity copper precursors is extremely challenging. The paper presents a comparison of analytical performances of the inductively coupled plasma optical emission spectrometry (ICP-OES) and the inductively coupled plasma mass spectrometry (ICP-MS) for the trace analysis of copper(II) acetylacetonate. A detailed study of the copper matrix effect and its influence on the intensity of the analytical signals was performed. The limits of detection for ICP-OES and ICP-MS analysis ranged from 0.1 to 100 ng g[-1] for more than 60 trace elements. The capabilities of the ICP-MS technique were expanded by the use of a reaction/collision cell. The combination of the developed ICP-OES and ICP-MS methods allows monitoring 62 impurities in copper(II) acetylacetonate with a purity of 6N (99.9999 wt%). The accuracy was rigorously validated through spike recovery experiments and the analysis of a high-purity copper certified reference materials, demonstrating the absence of significant systematic errors. The methodology's applicability was further confirmed by analyzing additional copper precursors, showcasing its utility for quality control of copper ALD materials.