Non-invasive optical quantification of methanol in bottled spirits

Abstract

Food and beverage contamination poses a persistent global threat. A prime example is the presence of methanol in counterfeit or illicit spirits, causing severe and often fatal poisoning worldwide. Rapid, non-destructive, and on-site screening methods capable of molecular analysis directly through commercial packaging are therefore urgently needed for quality control and consumer safety. Here, we introduce a non-invasive optical approach based on Raman spectroscopy that judiciously combines wavefront shaping with wavelength modulation to enhance the signal-to-noise ratio and enable quantification of methanol in unopened bottled spirits. A limit of detection of 0.2% (v/v) methanol in 40% ethanol was achieved, well below the 2% (v/v) threshold for safe human consumption. This truly non-invasive method remains robust through coloured glass bottles, with calibration validated in a real spirit sample. By enabling through-container methanol detection, the technique offers a practical tool to protect consumers and streamline routine screening across the beverage supply chain. Moreover, this Raman geometry establishes a versatile platform for assessing authenticity, composition, and contaminants directly through packaging.

Figures (17)

• Figure 1: Comparing RS with axicon and standard RS. Conceptual diagrams of the through-bottle Raman geometry using a) a conical excitation beam and b) a Gaussian excitation beam. Example through-bottle spectra of whisky (Whisky 4) acquired with c) the RS with axicon configuration and d) the standard RS configuration. The spectrum collected with the axicon geometry is dominated by the signal of the contents, giving a signal-to-background (SBR) of 0.42 for Peak A. In contrast, the standard RS spectrum is dominated by the fluorescence of the container, with only small Raman peaks visible (SBR = 0.10). For reference, c) and d) include a spectrum of the whisky measured directly through a vial (no bottle) and the spectrum of the empty bottle (Bottle 4), respectively. Raman spectra of Whisky 4 measured through different bottles (Bottles 1-4) using e) RS with axicon and f) standard RS. The axicon-based measurements are more consistent and primarily reflect the signal from the contents, while the standard RS spectra vary more due to the influence of the different containers. The spectra in e) and f) were normalised to the main ethanol peak (Peak A) to account for intensity variations caused by the different containers.
• Figure 2: WMRS principle and analysis. In WMRS, multiple Raman spectra are recorded while changing the wavelength of the excitation light ($\lambda_{\text{ex}}$). a) Seven Raman spectra of whisky (Whisky 4) were measured through its bottle as $\lambda_{\text{ex}}$ is varied over a range of 1 nm. For small wavelength changes, the Raman peaks shift accordingly while the background fluorescence stays constant. In b), a single Raman spectrum excited at the central wavelength ($\lambda_{\text{ex4}}$) is shown. c) Principal component analysis (PCA) performed on the seven spectra identifies a first principal component (PC1) with loadings which indicate the position and relative intensity of the Raman peaks. The Raman peak positions correspond to the zero-crossing of the peaks in the WMRS spectrum, and the relative Raman peak intensities are retained. Importantly, this new spectrum does not contain features from wavelength-independent fluorescence, meaning Raman peaks that are difficult to identify with standard RS (consider Peak C) become more evident with WMRS.
• Figure 3: Improved SNR in through-bottle Raman spectroscopy using wavefront shaping and wavelength modulation. The four methods (standard RS, RS with axicon, WMRS, and WMRS with axicon) were used to acquire the Raman spectra of commercial spirits through their sealed bottles. The relative SNRs obtained with each method for all the containers are represented in the bar graph, normalised to the SNR of the standard RS for clarity. Representative spectra obtained using the four configurations through two of the containers (Bottles 2 and 4) are displayed above. The inset is a photo of cut-outs (65 mm $\times$ 55 mm) of the bottles used in this experiment.
• Figure 4: Calibration graph for methanol detection. a) Standard Raman spectra and b) corresponding WMRS spectra of pure ethanol (EtOH) and methanol (MeOH) (left), and mixtures of 40% (v/v) ethanol with increasing methanol concentration (right). Zoomed-in insets highlight the spectral region used to extract the methanol peak height. At low concentrations, negative peak heights arise in the WMRS spectra from calculating the peak-to-peak distance (orange lines). c) Calibration regression curve showing a linear relationship between methanol peak height and methanol concentration, measured using the WMRS with axicon configuration. Each point is the average of five replicate measurements through Bottle 4; measurements through three additional bottles (Bottles 1 -- 3) are also shown. The LOD and LOQ are indicated on the graph, along with the legal EU limit and the maximum tolerable threshold for safe human consumption. Statistical significance between neighbouring calibration points was assessed with an unpaired t-test (* P$<$0.05, ** P$<$0.01, and *** P$<$0.001).
• Figure S1: A schematic of the system used to measure the Raman spectrum of spirits in a sealed bottle. A wavelength-tunable laser source, centred at 785 nm, is directed through an axicon lens to create a Bessel beam. This beam is relayed to the sample plane by lenses L2 and L1, resulting in an annular beam incident on the surface of the bottle before it focuses to a point inside the bottle. The Raman signal of the contents inside the bottle is collected in a backscattered configuration through the 'dark' centre of the excitation beam and directed to a spectrometer via a dichroic mirror. An iris is added in the collection path to eliminate the signal of the bottle that is excited by the incident annular beam. MMF: multi-mode fibre; DM: dichroic mirror; L1: lens (40 mm); L2: lens (100 mm) and L3: lens (50 mm).