Assessing Vodka Authenticity and Origin in Vietnam's Market: An Analytical Approach Using FTIR and ICP-MS with Multivariate Statistics

Vodka constitutes a significant sector of Vietnam's alcohol industry, including both domestic and imported varieties. However, this diversity faces challenges from illegal imports and adulterated products, threatening consumer health and brand integrity. This study employs Fourier transform infrared spectroscopy (FTIR) and inductively coupled plasma mass spectrometry (ICP-MS) to analyze 300 vodka samples from five brands collected across Hanoi. Significant variations were found in elemental compositions, with sodium concentrations ranging from 205.67 μg/L to 1269.24 μg/L and magnesium levels from 65.57 μg/L to 1453.34 μg/L. Principal Component Analysis (PCA) of the FTIR and ICP-MS data effectively differentiated the samples, with the first two principal components explaining 84.78% and 73.02% of the total variance, respectively. The PCA plots revealed distinct chemical profiles, notably isolating Rocket Vodka. These findings enhance food safety enforcement, protect consumer rights, and preserve brand reputations. The study underscores the importance of advanced analytical tools in combating beverage adulteration, ensuring public health, and maintaining market integrity, offering a replicable model for similar research in other regions.


Introduction
Food control and provenance determination have numerous advancements by using the fngerprinting technique in food authentication [1][2][3][4].Most of these researches were based on data obtained from analytical techniques and processed with one or multiple multivariate statistical analysis techniques [2,4,5].Te samples of interest are frst analyzed with a specifc analytical technique to achieve a data set which includes mostly quantitative information of elements or compound of interest such as traced element content, isotope ratio, or spectrum of the samples [5][6][7][8].Ten multivariate statistical analysis method will be applied to study the signifcant data and categorize the samples based on these signature diferences [9,10].Te methods rely on the assumption that certain components of the production conditions and environment will refect on the chemical composition of the product [5][6][7][8].
Vodka is a distilled alcoholic beverage, originating in Poland and Russia, consisting mainly of water and ethanol (37.5-55% alcoholic content) [11].Tis alcohol is the product of ethyl alcohol of agricultural origin such as the fermentation of potatoes, grains, or other agricultural products [12,13].Te characteristic of vodka comes from multiple fltering through activated charcoal, diluting with water, distilled, and demineralized [14][15][16].Traditionally, vodka is clear, colorless, with no aroma or taste, and the Russians called it voda for "water" [17].However, most brands in the vodka market now have both traditional and favor vodka.Te changes in the environment have brought changes to the distillery industry and the production of some type of vodka [13,18].Furthermore, with the standard properties of being clear, colorless, with no aroma and taste, an easy production method with wide range of base material, and the price range higher than beer, vodka becomes the target for illicit products [13,19,20].Te variety of vodka brands and types available on the market has become the perfect environment for adulterated and counterfeit vodka to thrive.
Drinking alcohol in Vietnam is widely considered a social etiquette, a way to build and maintain social networking, in both workplace and household etiquette.Also, Vietnamese consume alcoholic beverages for any celebration [21].Hence, Vietnam was ranked 16 th among the largest alcoholconsuming countries in 2016, ranked 2 nd in South East Asia, and 3 rd in Asia in 2022 [21,22].Te high consumption rate leads to high demand for products, and illicit products started to infltrate the alcohol market.According to Trend Economy, Vietnam's alcohol export in 2020 was up to 50% of total beverages export (over 121 million USD) [23].Currently, in Vietnam, many counterfeit products are being sold widely on the market; this status does not make an adverse impact on the health of consumers but also on the image, trademark, and economy of the manufacturer, causing loss on sales.Consuming adulterated/illicit alcohol can cause alcohol poisoning that leads to liver failure, kidney failure, blindness, coma, and eventually death; patients who survive alcohol poisoning can have hepatitis, gastrointestinal bleeding, cardiovascular disease, and high blood pressure [20,24,25].
Tere are reports of discriminating food products using methods such as gas chromatography (GC) and high performance liquid chromatography (HPLC) [15,26].However, such methods require complicated optimization and costly reference material [15,26].On the other hand, Fourier transform infrared-attenuated total refectance (FTIR-ATR) and inductively coupled plasma--mass spectroscopy (ICP-MS) are quick and efective methods, especially for liquid samples [4,[27][28][29].Terefore, this research aims to identify diferent vodka products available in Vietnam's market using ICP-MS and FTIR results combined with multivariate statistics.
Fourier transform infrared (FTIR) spectroscopy is based on the diferences in the structure of each molecule and they absorb a diferent amount of energy from the infrared (IR) source [30,31].Te covalent bonds within the molecule absorb specifc wavelengths and change the vibrational energy in the bond.Te vibrational type (stretching or bending) caused by IR radiation depends on the atoms in the bond [30,31].Attenuated total refection (ATR) is a method to directly observe a sample in the solid or liquid state.ATR involves an IR beam that travels from a medium with a high refractive index (ATR crystal) to a low refractive index (sample) and refects with reduced intensity [32].
Inductively coupled plasma mass spectrometry (ICP-MS) is a multielement analytical technique to determine trace element levels [27,33].Tis method is applied in many diferent felds such as environment (water, soil, air, sediment, and waste), product testing, quality control, agriculture, and food safety [27].

Sample Collection and Preparation.
Vodka samples from fve distinct brands-Vodka Hanoi, Men's Vodka, Aligator Vodka, Hanoi Wine, and Rocket Vodka-were systematically collected from major retail outlets and supermarkets across Hanoi.A total of 300 samples were acquired, with each brand contributing 60 samples.To ensure integrity and prevent contamination, each sample was maintained in a fully sealed condition at ambient room temperature.Furthermore, all samples were meticulously labeled with details including the location of collection, date of processing, and a unique coding number to facilitate precise tracking and identifcation during analysis.

IR Spectral Acquisition.
Te spectral acquisition was conducted using the Nicolet iS50 FTIR Spectrometer (Termo Fisher Scientifc, Massachusetts, United States), which is equipped with an integrated attenuated total refection (ATR) system featuring a zinc selenide (ZnSe) crystal.Tis setup allows for the measurement of both mid and long infrared regions.For the analysis, 3 to 4 drops of each sample were directly applied onto the ZnSe crystal surface.Each sample underwent 16 scans at a resolution of 4, and the procedure was repeated six times for each sample to verify reliability.A blank measurement was conducted between each sample analysis to ensure accuracy.Te spectrometer was meticulously cleaned with ethanol following each measurement session.Te spectral data were captured and analyzed using OMNIC software, covering a wavelength range from 4000 to 400 cm −1 .

Elemental Determination by ICP-MS
2.3.1.Chemicals.Nitric acid (HNO 3 ) at a 65% concentration and hydrogen peroxide (H 2 O 2 ) at 30% were sourced from Merck, USA.Ultrapure deionized water, possessing a resistivity of 18.2 MΩcm, was produced using the Milli-Q Plus water purifcation system (Millipore, Bedford, MA, USA).Certifed Reference Materials (CRMs) were employed from CPA Chem, which provided a standard solution containing 100 mg/L of multiple elements including Al, Ag, As, B, Ba, Be, Bi, Ca, Cd, Cs, Co, Cr, Cu, Fe, In, K, Li, Mg, Mn, Mo, Na, Ni, Nb, Pb, Rb, Sb, Se, Sr, Ti, Tl, V, U, and Zn, all in a 5% nitric acid solution.Tis was used to construct the calibration curves.Additionally, a solution of 9 elements at 10 mg/L concentration-Bi, Ho, In, 6 Li, Lu, Rh, Sc, Tb, and Y, also in a 2% nitric acid solution-was utilized as an internal standard for the analyses.

Sample Preparation and ICP-MS Methods.
Each vodka sample, initially 50 mL in volume, was heated at 80 °C until reduced to 20 mL.Tis concentration step primarily aims to remove ethanol, a volatile component that could create hazardous pressures during subsequent microwave digestion processes.After cooling, the concentrated sample was transferred to a 50 mL volumetric fask and diluted to the mark with ultrapure deionized water.For chemical 2 Journal of Analytical Methods in Chemistry analysis, 5 mL of this prepared solution was pipetted into a Tefon tube.To this, 2 mL of 65% nitric acid (HNO 3 ) and 0.25 mL of the internal standard were added.Te mixture was allowed to sit overnight to stabilize.Te samples were then subjected to microwave-assisted digestion using a MARS 6 system (CEM, North Carolina, United States) with the following digestion protocol specifcally designed for food samples: power set between 1030 and 1800 W, a ramp time of 20-25 minutes, a hold time of 15 minutes at 210 °C, followed by a cooling period of 20 minutes.Post-digestion, the digested solution was transferred to a 25 mL volumetric fask and brought up to volume with deionized water.Te fnal step involved fltering the prepared sample into a coded falcon tube, ensuring it was ready for inductively coupled plasma mass spectrometry (ICP-MS) analysis.Tis procedure ensures the removal of organic contaminants and the stabilization of the sample matrix, facilitating accurate trace elemental analysis.
Te experiment uses iCapTQ ICP-MS (Termo Fisher Scientifc, Bremen, Germany) to analyze the elements in the samples.Te elements and their most abundance isotopes measured are 11 B, 23 Na, 24 Mg, 39 K, 43 Ca, 48 Ti, 51 V, 52 Cr, 55 Mn, 57 Fe, 59 Co, 60 Ni, 63 Cu, 66 Zn, 98 Mo, 111 Cd, 115 In, 121 Sb, 138 Ba, 202 Hg, and 208 Pb.Te samples were diluted appropriately with deionized water and measured with the instrument.Te operational parameters for the ICP-MS analysis were as follows: RF power was set to 1200 W, with a sample depth of 5 mm.Te plasma gas fow rate was maintained at 15 L/min, the carrier gas fow rate at 1.05 L/min, and the makeup gas fow rate at 0.9 L/min.Te spray chamber temperature was controlled at 2 °C.For spectral analysis, each spectral peak was measured with 3 points per peak, and each reading consisted of 10 sweeps.Te samples were appropriately diluted with deionized water before measurement.
In this study, the instrument detection limits were calculated using the raw intensity data from the standard and the blank (using an ultrapure 2% nitric acid matrix) according to the equation as follows: in which IDL is the instrument detection limit, SD Blank is the standard deviation of measurement, C x is the mean signal for the standard, S x is the signal for C x , and S Blank : is the signal for blank.Te method of detection limit (MDL) was calculated as 2.4.Statistical Analysis.Statistical analysis of the collected data was conducted using STATISTICA 12 software (Dell, USA).Principal Component Analysis (PCA) was employed to explore and visualize the diferences among vodka brands based on data obtained from both Fourier transform infrared spectroscopy (FTIR) and inductively coupled plasma mass spectrometry (ICP-MS) analyses.Te results of this multivariate statistical technique were presented in two forms: a score scatter plot derived from FTIR analysis and a combined score scatter plot with X and R moving charts from the ICP-MS analysis.Tese visual representations allow for the identifcation of clustering patterns and outliers, thus highlighting the variances between the brands based on their spectral and elemental signatures.Tis analytical approach enhances the understanding of brand-specifc characteristics and assists in the authentication process.

Results and Discussion
3.1.Infrared Spectroscopy Analysis.Figure 1 illustrates a typical characteristic FTIR spectrum of a vodka sample, displaying various peaks across the spectrum.Despite initial observations, when the spectra from diferent brands are superimposed, the distinctions among them are minimal, indicating high similarity across the brands in the spectral data.Given the challenge in discerning diferences based solely on the full spectra, a targeted approach was adopted.Specifc, signifcant peaks within the spectrum were carefully selected for deeper analysis.Tese selected peaks were then analyzed using multivariate statistical analysis, specifcally Principal Component Analysis (PCA).Tis method enabled a more detailed examination of the underlying diferences and facilitated the classifcation of the vodka samples according to brand.Tis approach underscores the utility of combining targeted spectral analysis with advanced statistical techniques to enhance diferentiation and classifcation in complex sample sets.
In the analysis of vodka samples, specifc functional groups' peaks were identifed to correlate with the possible classes of compounds present in the beverage, as referenced in the literature [10] (Table 1).Among these, the peak at 3347 cm −1 , located within the range of 3600−3000 cm −1 , is attributed to the -OH (hydroxyl) group stretching, a characteristic feature of alcohols and phenols [32].Another signifcant band observed at 2980 cm −1 , falling within the 3000−2500 cm −1 range, is associated with the stretching vibrations of -CH groups, indicative of the presence of alkenes, alkynes, and alkanes [32,33].
Furthermore, a distinct sharp peak at 1640 cm −1 is identifed as the C�C stretching typically of monosubstituted alkenes [32].In the spectral region between 1300 and 1000 cm −1 , notable peaks include one at 1084 cm −1 , which is consistent with the C-O stretching vibrations found in alcohols.Additionally, a peak at 1043 cm −1 suggests the presence of S�O stretching, characteristic of sulfoxide compounds [32].Tese identifcations aid in the deeper chemical profling of vodka samples, providing insights into their compositional nuances.Te analysis reveals the presence of macro-elements such as Na ranging from 205.67 to 1269.24 μg/L, Mg from 65.57 to 1453.34 μg/L, K from 239.33 to 5856.33 μg/L, and Ca from 255.72 to 3157.16 μg/L.Te concentrations of these elements vary signifcantly between the brands, highlighting distinct profles.However, these macro-elements are also prone to environmental infuences due to their ubiquitous presence.For example, elements like Fe and Cu may derive from distillery equipment and processes [34], while Pb can be introduced through contaminated water supplies [35].Given these factors, the reliability of these elements as indicators of brand distinction or authenticity in vodka might be compromised.

ICP-MS Elemental Measurement.
To address this, the Principal Component Analysis (PCA) used for brand diferentiation and classifcation in this study selectively employs a subset of trace elements less likely to be infuenced by environmental factors.Te elements included in the PCA are Li, Al, Ti, V, Mn, Cr, Co, Ni, Zn, Mo, Cd, Sb, and Ba.Tese choices enhance the analytical rigor and reliability of the PCA results by focusing on elements with more controlled and specifc sources in the production process.

Brand Classifcation by Principle Component Analysis (PCA).
Principal Component Analysis (PCA) was applied to examine the compositional diferences among 300 vodka samples from fve brands, efectively visualizing their distinctiveness.Te PCA model utilized score scatter plots, which served as a screening tool to display the contributions of the principal components (PCs) to the overall variance observed within the sample dataset.Tese plots revealed that the selected PCs account for a signifcant proportion of the total variation, indicating that these components encapsulate the essential information of the variables under study.
Te efcacy and reliability of the PCA models are quantitatively assessed using the R 2 and Q 2 values.Te R 2 value measures the proportion of variance in the data that is explained by the PCA model, serving as an indicator of the model's ft.A higher R 2 value suggests a model that accurately captures the major trends and diferences among the samples.Conversely, the Q 2 value assesses the model's predictive ability based on a cross-validation procedure.It indicates how well the PCA model can predict new data points and is crucial for evaluating the model's generalizability and robustness.
By analyzing these values, researchers can determine the quality of the PCA model and its utility in distinguishing between vodka brands based on their chemical composition.Tis methodological approach provides a robust framework for identifying brand-specifc signatures and ensuring the authenticity of products in the market.

Principal Component Analysis of FTIR Data. Te outcome of the Principal Component Analysis (PCA) vividly
illustrates the compositional distinctiveness among the fve vodka brands (Figure 2).Te PCA results clearly demarcate each brand, although the distributions for Men's Vodka and   Tis graphical representation serves as a powerful tool for understanding how each brand's specifc compositional traits contribute to their placement on the PCA plot, thereby providing insights into the underlying variables that drive brand diferentiation in the vodka market.Such analysis not only aids in quality control and brand authentication but also helps in understanding consumer preferences linked to compositional diferences.
Table 3 in the PCA analysis of the FTIR data set indicates that the sum of squares of the frst four components accounts for 91.28% of the total variation in the samples.Notably, the frst two principal components (PCs) contribute signifcantly, representing 84.78% of this variation.Tis substantial proportion suggests that these initial two PCs encapsulate the primary information about the variables being analyzed.
Te efectiveness of the PCA model is quantifed by several key metrics: R 2 and Q 2 values.Specifcally, the cumulative R 2 X (R-squared) value for the frst two components stands at 84.78%.Tis R 2 X value refects the proportion of variance in the dataset that is successfully explained by PCs 1 and 2, confrming a strong model ft that captures the majority of data variability.Furthermore, the cumulative R 2 (Q-squared) value, which measures the model's predictive ability based on cross-validation, is recorded at 76.05% for the frst two components.
Te high Q 2 value indicates that the model not only explains but also reliably predicts new data within this variance scope.Together, these values demonstrate that the frst two principal components are highly efective in capturing and predicting the essential characteristics of the vodka samples based on FTIR data.Tis robustness in model quality underscores the utility of PCA in detailed compositional analysis, facilitating insightful distinctions and reliable predictions in complex datasets like those typical in quality control and brand diferentiation studies in the beverage industry.Journal of Analytical Methods in Chemistry

Principal Component Analysis of ICP-MS Data.
Te scatter plot derived from the PCA analysis provides a visual representation of the distribution of vodka samples based on their metal content (Figure 3).Notably, the plot reveals a partial overlap between the samples from Rocket Vodka and Men's Vodka.Tis overlapping suggests a degree of similarity in the metal composition of these two brands, indicating that they may share certain sourcing or production processes that afect their elemental profles.
Conversely, the samples from Hanoi Wine and Vodka Hanoi are distinctly separated on the plot, indicating a unique metal content that diferentiates them from each other and from the rest of the brands analyzed.Tis clear demarcation suggests that the metal constituents in Hanoi Wine and Vodka Hanoi are signifcantly diferent, which could be due to diferences in the raw materials used, the distillation process, or other factors specifc to each brand.
Such distinctions in the PCA scatter plot are crucial for identifying and understanding the underlying factors that contribute to the compositional diferences among brands.Tey not only help in the authentication of the vodkas but also in tailoring specifc quality control measures for each brand based on their unique metal content profles.
Table 4 from the PCA analysis on a diferent dataset reveals that the sum of squares for the frst eight principal components (PCs) accounts for 97.51% of the total variation among the vodka samples.Specifcally, the frst two components make up a substantial portion, contributing 73% to this total variance.Tis high proportion underscores the signifcance of these two components in capturing the essential characteristics and variability in the dataset.
Te quality of the PCA model is further quantifed by the cumulative R 2 X (R-squared) and Q 2 (Q-squared) values associated with these components.Te R 2 X value stands at 73.02%, indicating that approximately 73% of the variance in the sample data can be explained by the model constructed using just the frst two PCs.Tis suggests a strong model ft, capturing a large fraction of the information contained in the data.
Te cumulative Q 2 value is 67.48%, which measures the predictive accuracy of the model based on a cross-validation method.A Q 2 value of 67.48% implies that about 67% of the dataset's variance can be predicted by the model, highlighting its practical efectiveness in forecasting or simulating similar data.Tese metrics are essential for evaluating the PCA's capability not only to describe but also to anticipate the behavior of new sample data, which is particularly valuable for process control and quality assurance in vodka production.
Te two-way scatter plot in Figure 4, constructed using the frst two principal components (PC1 and PC2), provides a graphical representation of the infuence of various elements on the PCA results.In this plot, the position of each element relative to the origin of the coordinate system indicates its level of infuence on the analysis.Elements that are positioned further from the origin exert a more signifcant impact on the PCA model, refecting their substantial role in diferentiating the samples.
According to the plot, elements such as V, Ba, Li, Cr, Al, Mo, Zn, Ni, Sb, and Ti are identifed as having a heavy infuence on the PCA results.Teir strong presence on the plot suggests that these elements are key discriminators among the vodka samples, contributing signifcantly to the variance captured by the frst two principal components.
Tis visualization is crucial for understanding which variables (in this case, elements) are most critical in the classifcation and diferentiation process within the dataset.By identifying these infuential elements, researchers and quality control teams can focus on these specifc components during analysis, ensuring more targeted and efective monitoring and verifcation of the vodka samples.Tis approach not only enhances the robustness of the PCA   Te moving range charts provide a statistical depiction of the distribution of metal concentrations among various vodka brands (Figure 5).Tese charts are useful for visualizing both the central tendency (mean) and variability (range) of metal concentrations within the samples.
From the analysis, it appears that the brands Aligator, Men's, and Rocket Vodka exhibit relatively similar metal concentrations across all cases, suggesting a degree of uniformity in either their production processes or sources of raw materials.Such consistency might imply standardization or shared supply chains among these brands.
In contrast, Vodka Hanoi consistently shows the highest metal concentrations in seven out of the ten charts analyzed.Tis could indicate diferent sourcing of ingredients, distinct manufacturing processes, or less rigorous purifcation steps compared to the other brands.High metal concentrations might afect the taste, quality, and possibly safety of the vodka, depending on the specifc metals and their concentrations.
On the other hand, Hanoi Wine's vodka generally has the lowest metal concentrations in nine out of ten charts, except for vanadium, where it exhibits the highest concentration.Tis overall lower level of metals suggests more efective purifcation or diferent sourcing and production practices that minimize metal content, potentially leading to a cleaner and possibly more premium product.
Tese fndings highlight signifcant diferences in metal content between brands, which could be critical for consumer safety, regulatory compliance, and brand positioning in the market.Quality control measures and sourcing practices might need to be reviewed, especially for brands like Vodka Hanoi, to ensure product safety and maintain consumer trust.

Conclusion
Te illicit alcohol trade poses signifcant risks to human health and the global economy, making the development of robust quality assurance methods for legitimate alcoholic products critically important.Techniques such as multielement fngerprinting via inductively coupled plasma mass spectrometry (ICP-MS) and vibrational spectroscopy using Fourier transform infrared (FTIR) spectroscopy ofer rapid and efective means for analyzing the composition of liquid samples.Te application of multivariate statistical methods, particularly Principal Component Analysis (PCA), has proven efective in analyzing and categorizing extensive datasets.In the context of this study, PCA has successfully diferentiated 300 vodka samples based on brand distinctions, demonstrating its utility in distinguishing between products with high precision.
Tis capability highlights the potential to identify unknown vodka samples purely based on their FTIR or ICP-MS data profles.Such analytical techniques can be integrated into quality assurance protocols to authenticate vodka brands and detect counterfeits, thereby safeguarding consumer health and protecting brand integrity.Furthermore, these methods ofer a scalable approach to tracking and monitoring vodka products in the marketplace, ensuring compliance with safety standards and helping to combat the trade in illicit alcohol.Tis strategic application not only enhances consumer trust but also supports regulatory bodies in enforcing food and beverage safety regulations.

Figure 1 :
Figure 1: FTIR spectrum of a vodka sample.

4
Journal of Analytical Methods in ChemistryHanoi Wine are relatively proximate to each other.Despite their closeness, a discernible boundary still separates these two brands, suggesting subtle but signifcant diferences in their compositions.Te PCA plot further highlights that Rocket Vodka's results are positioned distinctly in the third quadrant, isolating it from the clusters formed by the other brands.Tis spatial separation within the PCA plot indicates that Rocket Vodka possesses a unique chemical profle compared to the others, emphasizing the efectiveness of PCA in capturing and visualizing such diferences.

Figure 2 :
Figure 2: PCA score plot of FTIR analysis of vodka samples.

Figure 3 :
Figure 3: PCA score plot of ICP-MS analysis of vodka samples.
Table 2 presents the elemental composition of vodka from fve diferent brands: Hanoi Wine, Men's Vodka, Aligator Vodka, Vodka Hanoi, and Rocket Vodka.

Table 1 :
Corresponding functional groups and possible compound class of FTIR signal of vodka sample.

Table 2 :
Te average content of 21 elements in vodka from 5 brands (μg/L).

Table 4 :
ICP-MS Principal Component Analysis summary.

Table 3 :
FTIR Principal Component Analysis summary.
6Journal of Analytical Methods in Chemistry model but also aids in pinpointing potential areas for quality improvement or regulatory focus in vodka production.