Infrared imaging identifies rancid hazelnuts without opening the bag
A hyperspectral imaging method developed by researchers at Universitat Rovira i Virgili (URV) allows producers to detect rancid hazelnuts inside sealed plastic packaging – without the need for destructive sampling. Using near-infrared hyperspectral imaging (NIR-HSI), the system tracks lipid oxidation, the chemical process behind rancidity, by detecting changes in the spectral profile of hazelnuts over time.
The method was tested across 78 days of storage under varying light and atmospheric conditions, revealing that oxidation rates depend most heavily on time, but are significantly influenced by oxygen availability and light exposure.
"The hyperspectral camera is here to stay," said lead researcher Jokin Ezenarro in a URV press release. Vacuum-sealed nuts exhibited the least oxidation, while those exposed to light and air degraded fastest. Sensory testing confirmed that spectral signs of degradation correlated with rancid taste and texture, validating the technique as a non-invasive tool for real-time quality monitoring.
Mid-infrared sensor tech goes compact and CMOS-compatible
Researchers at KAIST have developed a room-temperature mid-infrared photodetector that could accelerate the miniaturization and mass production of optical sensors for environmental, medical, and aerospace applications. Unlike traditional systems that require cryogenic cooling, the new device operates without cooling and is compatible with standard silicon-based CMOS fabrication, offering low-cost scalability and a compact footprint.
The photodetector integrates a germanium-based waveguide and leverages the bolometric effect – where light-induced heating causes electrical resistance changes – to detect a broad mid-infrared spectrum (4030–4360 nm). It demonstrated real-time CO₂ gas detection on a single chip and achieved record-high responsivity among waveguide-integrated thermal detectors.
“This research represents a novel approach that overcomes the limitations of existing mid-infrared photodetector technologies,” said lead researcher Professor SangHyeon Kim in a KAIST release.
Machine learning sharpens soil contaminant detection
Researchers at Rice University and Baylor College of Medicine have developed a new machine learning–enabled method for detecting toxic soil pollutants like polycyclic aromatic hydrocarbons (PAHs) – even when no experimental reference samples exist. The technique pairs surface-enhanced Raman spectroscopy (SERS) with in silico spectral libraries created via density functional theory (DFT), and uses physics-informed algorithms to compare spectral “fingerprints” with virtual references.
The approach sidesteps the need for physical standards by using two algorithms: characteristic peak extraction (CaPE) and similarity scoring (CaPSim), which together identify target compounds in complex soil matrices with high accuracy.
“This method makes it possible to identify chemicals that have not yet been isolated experimentally,” said study co-author Naomi Halas in a Rice University release. Tested on spiked soil samples and validated against gas chromatography–mass spectrometry, the system shows promise for scalable, field-deployable environmental monitoring.
Single-molecule calorimetric force spectroscopy reveals DNA stability at base-pair level
Researchers at the University of Barcelona have used calorimetric force spectroscopy – a single-molecule approach combining optical trapping and temperature control – to measure how DNA strands separate, revealing detailed thermodynamic information at the resolution of individual base pairs. By unzipping a DNA hairpin while varying temperature, the team captured base-pair-specific changes in entropy, enthalpy, and heat capacity (ΔCp), key parameters that influence DNA stability but are often oversimplified in bulk melting experiments.
Unlike conventional methods, this technique directly links mechanical force measurements to thermal behavior, enabling a more nuanced understanding of DNA hybridization under varying conditions. The findings could refine predictive models used in molecular biology and biotechnology, particularly where precise control over DNA folding and stability is essential.
Newsletters
Receive the latest analytical science news, personalities, education, and career development – weekly to your inbox.
