**Background:** Wine composition analysis is critical for assessing quality, origin, and authenticity. Organic acids, phenolic compounds, amino acids, and metallic ions all contribute to wine flavor, aroma, and health-related properties. While mass spectrometry (MS) is widely used for wine analysis, most MS techniques can measure either organic or inorganic components but not both simultaneously. In situ liquid secondary ion mass spectrometry (SIMS), developed by the authors' group, can provide both organic and inorganic molecular information from liquid samples without prior preparation. Principal component analysis (PCA) is a powerful tool for extracting meaningful patterns from complex SIMS spectra containing hundreds of peaks.
**Methods:** Three batches of wines were analyzed. Batch 1 comprised six wines from Chateau Ste. Michelle Winery (Columbia Valley, WA, USA): three red wines (Cabernet Sauvignon 2016, Merlot 2016, Syrah 2017) and three white wines (Chardonnay 2017, Sauvignon Blanc 2017, Riesling 2018). Batch 2 consisted of four Cabernet Sauvignon wines from different producers/locations: Chateau Ste. Michelle (WA, 2018), Rutherford Ranch (Napa Valley, CA, 2018), Dreaming Tree (Chile, 2020), and Yellow Tail (Australia, 2021). Batch 3 comprised four Chardonnay wines: Chateau Ste. Michelle (WA, 2020), Barnard Griffin (WA, 2020), Bread and Butter (Napa Valley, CA, 2020), and Yellow Tail (Australia, 2020). All wines were purchased from Walmart in Richland, Washington. In situ liquid SIMS was performed using a ToF-SIMS V instrument (IONTOF GmbH, Münster, Germany) with a 25 keV pulsed Bi₃⁺ primary ion beam focused to ~400 nm diameter. A vacuum-compatible microfluidic device with a SiN membrane window was used to introduce liquid wine samples. Both positive and negative ion spectra were collected. Mass calibration used H⁺, CH₃⁺, OCH₃⁺, and Bi⁺ for positive mode and C⁻, C₂⁻, C₃⁻, and C₄⁻ for negative mode. The mass resolving power was 5,000–10,000, but only unit mass resolution was achieved with a 60 ns pulse width. Peak assignments were accepted if the relative mass deviation was below 300 ppm. Spectral datasets were normalized to total ion counts, square-root transformed, and analyzed by PCA using MATLAB. For comparison, traditional ToF-SIMS was performed on dried wine samples (10 μL deposited on clean Si wafers and air-dried).
**Key Results:** In negative ion mode, red wines showed higher normalized intensities of benzene-ring-related species (C₄H⁻ at m/z 49, C₄HO⁻ at m/z 65) and phenolic acids (gallic acid m/z 169, caffeic acid m/z 179, citric acid m/z 191, quinic acid m/z 191, ferulic acid m/z 193). White wines were enriched in non-phenolic acids such as malic acid (m/z 133) and tartaric acid (m/z 149). PCA of negative ion spectra separated red from white wines along PC1, with red wines having higher PC1 scores. The PC1 loadings showed positive loadings for benzene-ring species and negative loadings for malic, tartaric, and acetic acids. White wines were further separated along PC3, with Chardonnay 2017 containing more N-organic species, lactic acid, and HSO₄⁻, while Sauvignon Blanc 2017 contained more POₓ⁻ species and malic acid. In positive ion mode, red wines showed higher PC1 scores and contained more K⁺, N-related organic species, and high-mass organics, while white wines contained more oxygen-related organics and calcium. Anthocyanins (cyanidin, delphinidin, pelargonidin, peonidin, malvidin, petunidin) were detected as positive PC1 loadings in red wines. All six wines from the single manufacturer were well separated in the positive-ion PCA scores plot. For the four Cabernet Sauvignon wines from different locations, PCA of positive ion spectra (PC2 vs. PC3) achieved reasonable separation, with metal ions (K⁺, Mg⁺, Al⁺, Si⁺) and N/O-containing organics as key discriminating features. For the four Chardonnay wines, PCA of positive ion spectra (PC1 vs. PC2) effectively separated all four wines, with the Australian wine (Yellow Tail) clearly distant from the three US-produced wines. In contrast, traditional ToF-SIMS of dried samples showed K-salt species dominating the spectra (e.g., K₂H₂PO₄⁺ at m/z 175, K₃HPO₄⁺ at m/z 213), and PCA separation was less clear than with in situ liquid SIMS. Additionally, anthocyanin signals from dried samples contradicted known patterns (white wines appeared to contain more petunidin), indicating that drying alters sample composition.
**Clinical Implications:** This study does not have direct clinical implications. However, the analytical method described could be applied to quality control and authentication of wines and other food liquids (e.g., juice, milk, coffee), which may indirectly impact public health by ensuring food safety and authenticity. The ability to simultaneously detect organic and inorganic species without sample preparation represents a technical advance in food science.