Skip to content
Frontiers
Search

Brief Research Report

Frontiers in Natural Products ·

Volume 5 - 2026 | https://doi.org/10.3389/fntpr.2026.1886822

Physicochemical characterization and metabolomic analysis of green coffee oil from Coffea arabica L. cultivated in Bahia, Brazil

  • Jayne Ferreira da Silva Oliveira1
  • Paulinne Moreira Lima1
  • Maria Alice Sousa Marinho1
  • Micaela Sousa Coutinho Silva1
  • Lara Beatriz de Andrade Silva1
  • Isabela Dias dos Santos1
  • Bruna Santana Neves1
  • Anderson Santos Souza1
  • Gabriel Azevedo de Brito Damasceno1
  • Juliano Geraldo Amaral1*corresponding author
  1. Instituto Multidisciplinar em Saúde, Universidade Federal da Bahia, Vitória daConquista, Bahia, Brazil
Article metrics0 viewsView details
PDFCiteMetrics

Abstract

Vegetable oils have extensive industrial and food applications, with green coffee oil standing out due to the presence of bioactive compounds of pharmaceutical and cosmetic interest. This study aimed to characterize the metabolomic and physicochemical profile of oil extracted from Coffea arabica L. seeds cultivated in the southwestern region of Bahia, Brazil. Ten green coffee samples were subjected to oil extraction and analyzed for density, refractive index and acidity index, in addition to metabolomic analysis based on molecular networking. The results demonstrated the refractive index values between 1,473 and 1,478, density values between 0.9099 and 0.9851 and acidity index values ranging from 3 to 19 mg KOH/g. Metabolomic characterization enabled the identification of palmitic, linoleic, oleic, and stearic acids as the major fatty acids, along with other metabolites associated with the lipid fraction of the oil. Differences in the abundance of these compounds were observed among samples, highlighting chemical variability related to seed origin and characteristics. The predominance of bioactive fatty acids and compounds with antioxidant potential reinforces the applicability of green coffee oil as a promising raw material for cosmetic formulations and the development of pharmaceutical products.

Keywords

1. Introduction

Vegetable oils play a fundamental role due to their high demand and broad applications in the industrial and food sectors, being of great importance to human health due to the presence of bioactive compounds. Each vegetable oil possesses a unique combination of properties, with coffee oil standing out as a byproduct derived from one of the most widely consumed commodities worldwide. The genus Coffea occupies a singular position in socioeconomic history and phytochemical research. In addition to being one of the most commercially traded agricultural products worldwide, the different fractions of the fruit (pulp, husk, green, and roasted seeds) have proven to be rich sources of bioactive compounds with pharmacological and cosmetic potential (Makiso Urugo et al., 2023; Muñoz et al., 2020). This rediscovery of coffee as a functional raw material has increased interest in seed extracts and oils, making them the focus of recent phytochemical studies (Piasek et al., 2026; Guia do Café, 2024; dos Santos et al., 2021).

In Brazil, the world’s largest coffee producer, the cultivation of Coffea arabica L. is widely distributed geographically. Particularly in the southwestern region of Bahia, although not officially recognized among Brazil’s main traditional coffee-producing regions, this area stands out for the production of Arabica coffee cultivated under high-altitude and mild-climate conditions, factors known to influence the physicochemical and sensory attributes of coffee beans (EMBRAPA. Produção de café arábica, 2022). In 2022, approximately 20 tons of coffee beans produced in the municipality of Barra do Choça were exported, with an annual production of around 280 thousand bags, highlighting the growing importance of Barra do Choça coffee in the agricultural sector (BARRA DO CHOÇA, 2023). Similarly, Vitória da Conquista possesses all the attributes required for excellent Arabica coffee production (Pereira, 2018). Despite the regional agronomic relevance, studies focused on the characterization of oil extracted from locally produced green coffee seeds remain scarce, especially those employing an integrated metabolomic approach.

From a phytochemical perspective, the oil extracted from green C. arabica L. seeds presents a complex composition, containing triacylglycerols, free fatty acids, diterpenes such as cafestol and kahweol, lipid esters, phenolic acids, and alkaloids, including caffeine in smaller amounts (Makiso Urugo et al., 2023; Wang et al., 2025). Its lipid fraction is predominantly composed of mono- and polyunsaturated fatty acids, which also act as precursors of biologically relevant lipophilic diterpenes (Zhu et al., 2023). Phytochemical and pharmacological studies have demonstrated that these constituents are associated with antioxidant, anti-inflammatory, and antimicrobial activities, reinforcing the potential of green coffee oil as a source of bioactive compounds of pharmaceutical and cosmetic interest (Amiri et al., 2024; Ruse et al., 2025; Lima et al., 2025).

In addition to the intrinsic composition of the species, regional factors exert a significant influence on the chemical profile of primary and secondary metabolites in plants. Variables such as altitude, water availability, temperature, and soil characteristics may modulate biosynthetic pathways, directly impacting their composition. Therefore, chemical characterization associated with geographic origin becomes an essential tool for understanding metabolic variations and establishing quality standards (Zheng et al., 2023).

Traditionally, the quality of vegetable oils is evaluated through physicochemical parameters such as acidity index, density, and refractive index, which provide information regarding stability, degree of hydrolysis, and structural integrity of the compounds present (Herculano et al., 2021). However, these analyses do not allow a comprehensive elucidation of the molecular diversity present. In this context, gas chromatography coupled with mass spectrometry (GC–MS) stands out as a robust technique for the characterization of fatty acids and volatile compounds. In addition, approaches based on molecular networking, such as the Global Natural Products Social Molecular Networking (GNPS) platform, have expanded the capacity for exploration of spectrometric data, enabling structural organization, annotation, and visualization of chemical relationships among compounds (Wang Mingxun et al., 2016).

Despite advances in the chemical characterization of coffee beans and their byproducts, a gap remains in the literature regarding the integration of physicochemical parameters and spectrometric data. In this context, the present study aims to perform the extraction and metabolomic analysis of oil obtained from green coffee beans, as well as to evaluate its physicochemical properties. Furthermore, this study seeks to establish scientific foundations for the application of this oil, contributing to value addition within the regional production chain and fostering technological innovation.

2. Materials and methods

Plant material, extraction, and sample preparation

Ten independent lots of green Coffea arabica beans (SisGen: A4142D9) were obtained from different producers located in the municipalities of Barra do Choça and Vitória da Conquista, southwestern Bahia, Brazil. A total of 10 samples were analyzed, each consisting of 250 g of carefully selected seeds to ensure homogeneity. The beans were initially dried in a forced-air oven at 40 °C for 3 h to reduce moisture content and subsequently ground using a knife mill until a particle size of <300 nm was achieved. From each sample, 100 g of the powdered material was subjected to ultrasound-assisted maceration extraction (15 min/day for seven consecutive days), using hexane as the extraction solvent. The extraction conditions had been previously optimized through pilot-scale experiments. After extraction, the material was vacuum-filtered, and the solvent was removed using a rotary evaporator under controlled temperature conditions.

Physicochemical analysis

The acidity index was determined according to an adapted Lutz (LUTZ, 2008) methodology, using 0.5 g of oil sample dissolved in an ethanol:diethyl ether solution (2:1, v/v) containing phenolphthalein, followed by titration with 0.1 M NaOH until the endpoint was reached. The refractive index was determined using a refractometer calibrated at 20 °C, while density was measured using a calibrated pycnometer according to the Brazilian Pharmacopoeia (7th ed.) (Agência Nacional de Vigilância Sanitária ANVISA, 2023), by comparing the mass of the oil and distilled water at the same volume.

Chromatographic analysis

The phytochemical profile of the oil was analyzed by gas chromatography coupled with mass spectrometry (GC–MS), with the methodology described in the Supplementary Material. The compounds were annotated by comparison with the Wiley and NIST libraries, complemented by gas-phase fragmentation studies (Barth et al., 2015). The data were also processed using the GNPS platform, allowing the construction of molecular networks and the annotation of related chemical clusters (Wang Mingxun et al., 2016). The use of the GNPS platform aimed to complement the compound identification achieved by GC–MS analysis. While spectral libraries enabled the individual annotation of metabolites, GNPS allowed the organization of the data into molecular networks based on the similarity of fragmentation spectra. Furthermore, molecular networking facilitated the metabolomic interpretation of the results by highlighting associations between specific groups of compounds and ranges of density, refractive index, and acidity index.

Statistical analysis

Statistical analyses were performed using GraphPad Prism software (version 10.4.2), which was employed both for the regression model applied to the optimization of the extraction process and for the descriptive analysis of physicochemical parameters, including calculation of the mean, coefficient of variation, and Pearson’s correlation. Additionally, Principal Component Analysis (PCA) was conducted using Statistica software (version 13.5.0.17), based on data obtained from molecular networking through the GNPS platform, aiming to identify clustering patterns and discriminant variables among the samples.

3. Results

Optimization of the extraction method

Initially, optimization of the extraction method was performed using one of the green coffee samples, considering the number of days required to obtain the highest extraction yield as the variable (Supplementary Table S1). Linear regression analysis indicated a positive relationship between solvent contact time and oil extraction yield. The fitted model was described by the equation Y = 0.1365X + 1.421, with a coefficient of determination of R 2 = 0.8901. The slope coefficient presented a value of 0.1365, with a 95% confidence interval ranging from 0.08134 to 0.1916, demonstrating a progressive increase in yield with increasing contact time. Regression significance analysis showed that the slope was statistically different from zero (p = 0.0014), confirming the existence of a significant linear relationship between the evaluated variables. Based on these findings, the longest extraction period was adopted as the experimental condition for obtaining the oil samples used in the subsequent physicochemical and metabolomic analyses.

Density analysis

The density of the analyzed samples (Supplementary Table S2) presented values compatible with those described for vegetable oils, reflecting the predominant composition of triacylglycerols present in the lipid matrix. The slight variations observed among the samples may be related to differences in the fatty acid composition. Statistical analysis revealed low variability among the samples, with a mean value of 0.9503 g/cm 3 and a coefficient of variation of 3.09%, indicating stability and good analytical reproducibility.

Refractive index analysis

The refractive index values obtained (Supplementary Table S2) were within the expected range for vegetable oils, as this parameter is directly influenced by the chemical composition of the lipid fraction. In particular, the degree of unsaturation of the fatty acids present in the oil plays a relevant role in the variation of this parameter. The obtained values showed a mean of 1.475 and a coefficient of variation of 0.10%, indicating symmetric distribution and high analytical precision.

Acidity index analysis

The values obtained for the acidity index (Supplementary Table S2) demonstrated levels higher than those expected for vegetable oils, particularly coffee oil. Elevated values of this parameter may be associated with triacylglycerol hydrolysis processes, which may occur during seed storage, processing, or handling stages. The acidity values showed a mean of 12.86 mg KOH/g and a coefficient of variation of 40.20%, indicating high variability among the samples. These results suggest significant differences in the degree of triacylglycerol hydrolysis among the analyzed samples.

Pearson correlation analysis of physicochemical parameters

Pearson’s correlation analysis was performed to evaluate the relationship between the acidity index and the physicochemical parameters density and refractive index in the coffee oil samples (Supplementary Table S3). The Pearson correlation analysis did not reveal any significant association between the acidity index and the evaluated physicochemical parameters. The refractive index showed a very weak positive correlation with the acidity index (r = 0.0589; p = 0.8714), while density also exhibited a very weak positive correlation (r = 0.1172; p = 0.7472).

Chromatographic analysis

Gas chromatography analysis revealed intense peaks corresponding to palmitic acid and linoleic acid in all analyzed samples, as shown in Supplementary Table S4. Oleic acid, octadecanoic acid, and eicosanoic acid were annotated at lower intensities, in addition to β-tocopherol, α-tocopherol, stigmasterol, and β-sitosterol. Compounds derived from sterols and cafestol/kahweol were also annotated. Caffeine was annotated at low intensity.

Molecular networking

Regarding the constructed molecular network, the samples exhibited a similar relative abundance pattern for palmitic acid. In contrast, linoleic acid showed higher intensity in samples 6 and 8, indicating a greater relative abundance of this compound in these samples. Additionally, the nodes associated with cafestol/kahweol derivatives and sterols presented higher intensity in samples 8 and 10, suggesting a greater contribution of these metabolites to the chemical composition of these samples (Figure 1; Supplementary Figure S1).

Molecular network associated with physicochemical parameters

Acidity

The molecular network (Figure 2; Supplementary Figure S2) indicates clusters of structurally related metabolites. Fatty acids, such as palmitic acid, showed a balanced distribution, suggesting that this compound is not exclusively associated with a single acidity condition. In contrast, although linoleic acid contributed across all acidity ranges, its higher abundance was evident in samples presenting acidity values within the 15–20 range. A similar behavior was also observed for oleic acid and octadecanoic acid. Nevertheless, the distribution of the nodes demonstrated a predominance of samples within the 0–5 and 5–10 acidity ranges throughout most of the network, suggesting the presence of compounds associated with lower acidity indices, and consequently, lower levels of degradation.

Refractive index

Molecular network analysis (Figure 3; Supplementary Figure S3) revealed a predominance of the refractive index ranges 1.473–1.474 and 1.475–1.476 throughout the main clusters. Fatty acids such as palmitic acid, linoleic acid, oleic acid, and octadecanoic acid were distributed across all ranges, with a slight predominance in the lower range. In general, the different classes of annotated compounds showed greater distribution in the lower refractive index ranges, suggesting greater production of compounds at lower index values. The highest range (1.477–1.478) exhibited lower frequency and a more localized distribution.

Density

The molecular network revealed a predominance of the higher density ranges, especially within the 0.970.98 and 0.981.00 ranges, which were broadly distributed throughout the nodes present in the network (Figure 4). Palmitic acid, linoleic acid, oleic acid, and octadecanoic acid showed greater abundance in the higher range, whereas other classes, such as sterol derivatives, exhibited greater distribution within the lower intermediate range. The lowest density range (0.930.94) presented a more dispersed and localized occurrence, suggesting lower production of the different classes of compounds at lower density values.

Principal Component Analysis (PCA) of annotated fatty compounds and physicochemical parameters

Following molecular networking analysis, PCA) was performed to verify the existence of clustering patterns among the samples. For this analysis, 10 green coffee oil samples and eight variables (fatty acids and physicochemical parameters) were used, based on autoscaled data. The results obtained from the analysis are presented in Supplementary Figure 5. Before visualizing the score and loading plots, it is necessary to define the number of principal components (PCs) that best characterize the original dataset. PCs are new axes generated from the original variables. In the present study, the first two PCs were selected because together they accounted for approximately 70% of the cumulative variance of the data.

4. Discussion

Extraction optimization

The initial factors required to ensure optimized oil extraction involve a series of conditions. Among these, seed quality is one of the most critical aspects, as well-selected, mature, and contaminant-free seeds are essential for preserving the bioactive compounds and desirable characteristics of the oil. Coffee beans with high antioxidant and essential fatty acid contents, as well as a low incidence of damage, provide an adequate starting point (Barth et al., 2015; Tsukui et al., 2014). Another essential factor is particle size after the grinding process. Particles that are too large may result in low extraction efficiency, whereas excessively fine particles may hinder oil separation from the solid residue and increase the risk of thermal degradation (Baldelli and Aguilera, 2025; Thapa et al., 2025).

Furthermore, the extraction method may be a determining factor in oil quality. One of the most widely applied methodologies in extraction processes is solvent extraction, particularly using hexane, an alkane hydrocarbon commonly employed due to its physicochemical characteristics that meet the required extraction conditions (Brondani Teixeira Ribas et al., 2025a; Gebreegziabher et al., 2025; Popova and Bankova, 2023). Hexane readily interacts with the oil present in the material and possesses a low boiling point, an important feature for preventing oil decomposition. This technique is frequently used across different sectors, ranging from product manufacturing to oil extraction (Gebreegziabher et al., 2025).

During the extraction process, the lipid material gradually dissolves into the solvent through diffusion, a process dependent on concentration gradients and contact time (Ramos et al., 2024). The longer the hexane remains in contact with the seeds, the greater the amount of oil extracted, until equilibrium is reached. Additionally, the oil is not superficially available but rather located within the cellular structures of the seeds (Wang et al., 2025). Therefore, hexane must penetrate these structures in order for the oil to be solubilized, and this process occurs slowly. Increased contact time allows hexane to reach deeper regions of the plant material, resulting in higher oil recovery, as evidenced during the optimization of the extraction method (Brondani Teixeira Ribas et al., 2025b).

Linear regression analysis demonstrated a positive and significant relationship between solvent contact time and oil extraction yield, indicating that increasing extraction time favors the recovery of the lipid fraction from the plant matrix. The high coefficient of determination observed (R 2 = 0.8901) indicates that a large proportion of the variability in extraction yield can be explained by the time variable, suggesting that the extraction process is strongly dependent on the duration of solvent–sample contact. This behavior is consistent with solid–liquid extraction processes in plant materials, in which prolonged contact time enhances the diffusion and solubilization of lipophilic compounds present within plant cells, promoting greater mass transfer to the solvent. Work es on the extraction of lipids and metabolites from plant matrices have demonstrated that extraction yield tends to increase progressively over time until the system approaches extraction equilibrium, a behavior widely described for diffusional processes involved in the recovery of bioactive compounds (Brondani Teixeira Ribas et al., 2025b; Lin et al., 2025; Schoss and Glavač, 2024).

Physicochemical analyses

The refractive index measures the ratio between the speed of light in a vacuum and in a given medium, being influenced by factors such as the degree of unsaturation, oxidation, thermal treatment, and free fatty acid content, as highlighted by Lutz (LUTZ, 2008). Essay conducted by Oliveira (Maressa, 2013), based on 10 assays, reported refractive index values ranging from 1.4755 to 1.4810 at 25 °C, while Dantas (Dantas, 2017), evaluating coffee oil, found a refractive index of 1.482 at 26.5 °C. The results obtained in the present work (CV = 0.10%) are consistent with the literature data, reinforcing the consistency of the observed characteristics (Supplementary Table S2). The similarity among these values suggests a direct association with the unsaturation profile and other comparable physicochemical properties among the analyzed oils, thereby validating the reliability of the method and analytical conditions employed (Genovese et al., 2025; Dantas, 2017; Maressa, 2013).

Green coffee oil generally presents density values ranging from 0.91 to 0.94 g/cm 3, attributed to the presence of compounds such as fatty acids, which play a crucial role in its chemical composition and functionality (Pereira et al., 2016). Changes in density may indicate the addition of non-natural substances or the removal of essential components, reflecting possible modifications or adulterations. In the present study, the observed density values (CV = 3.09%) were consistent with data reported in the literature (Supplementary Table S2). Pereira (Pereira et al., 2016) reported a density of 0.91 g/cm 3, while Dantas (Dantas, 2017) described a value of 0.9195 g/cm 3. This similarity suggests that the physicochemical characteristics of the analyzed oil, such as the proportion of fatty acids and other constituents, were preserved and remained within the expected standards. The stability of these characteristics reinforces the integrity of the material.

The acidity index reflects the amount of free fatty acids present in the oil, which may increase due to hydrolysis or oxidation processes influenced by factors such as extraction method, storage conditions, and handling (Gu et al., 2024). Previous studies have reported a wide variation in acidity index values for coffee oils. Pereira (Pereira et al., 2016) reported an acidity index of 1.16 mg KOH/g, whereas Dantas (Dantas, 2017), using the same methodology, obtained even higher values of 13.74 mg KOH/g. Another work conducted by Constantini (Constantini et al., 2023) reported acidity index values of 2.244 mg KOH/g (Dantas, 2017; Pereira et al., 2016; Constantini et al., 2023). In the present study, the acidity values obtained (CV = 40.20%) were higher than those reported in the literature, which may indicate the presence of elevated levels of free fatty acids. This finding suggests possible alterations resulting from oil handling or unfavorable storage conditions. Furthermore, variations in the extraction method or increased exposure to oxidative processes may also be associated with the results obtained (Gharby et al., 2025; Galarza and Figueroa, 2022).

Pearson correlation analysis demonstrated that the acidity index was not significantly associated with either density or refractive index in the green coffee oil samples. The low correlation coefficients observed for refractive index (r = 0.0589; p = 0.8714) and density (r = 0.1172; p = 0.7472) indicate that variations in the acidity index occur independently of these physical properties. This behavior may be explained by the fact that the acidity index is directly related to the concentration of free fatty acids formed through lipid hydrolysis and degradation processes, whereas density and refractive index are mainly influenced by the overall composition of the lipid fraction, the degree of unsaturation, and the molecular distribution of triacylglycerols (Khaled et al., 2024; Correa et al., 2024; Zidane et al., 2026). Therefore, changes in free fatty acid content are not necessarily sufficient to produce detectable alterations in these physical parameters, which may explain the lack of correlation observed among the evaluated variables.

Chromatographic analysis and molecular networks

The analysis performed revealed the predominant production of palmitic acid and linoleic acid. This lipid profile is consistent with the composition of Coffea arabica L. oil, in which linoleic and palmitic acids represent the major constituents of the fatty acid fraction (Wang et al., 2025; Galarza and Figueroa, 2022). Linoleic acid, a polyunsaturated fatty acid, has been associated with anti-inflammatory and antimicrobial properties, as well as with the maintenance of skin barrier function, being widely used in cosmetic formulations. In contrast, palmitic acid, although a saturated fatty acid, contributes to the oxidative stability of the lipid matrix and may exert antimicrobial activity. The combination of these fatty acids, together with other metabolites present in coffee oil, may contribute to the attributed biological activities, reinforcing its potential application in pharmaceutical and cosmetic products (Herculano et al., 2021).

Based on the obtained mass spectra, the data were submitted to the GNPS platform. GNPS consists of a mass spectrometry-based algorithm that assists in the identification and characterization of substances through spectral libraries (Wang Mingxun et al., 2016; Wang M. et al., 2016). Thus, the samples demonstrated intense production of palmitic acid and linoleic acid, in addition to compounds derived from cafestol/kahweol and sterols, although at varying levels among the samples. In contrast, the low abundance of caffeine is attributed to its predominantly polar structural nature, which limits its presence in large quantities within the oil fraction (Tidke et al., 2024).

The differences in compound production among the samples are directly related to their physicochemical profile. Several factors, such as temperature, seed maturation time, harvesting, storage, and seed processing conditions, may affect the composition of the oil (Lima et al., 2025). It is well established that the biosynthesis and accumulation of lipid metabolites in plant seeds are highly sensitive to the environmental and physiological conditions of the plant. During seed development and maturation, intense metabolic activity occurs, in which enzymes involved in the biosynthetic pathways of fatty acids and lipids regulate the formation and relative proportion of the different fatty acids present in the oil. Changes in factors such as temperature and water availability may alter the activity of these enzymes, resulting in modifications in the degree of saturation, carbon chain length, and relative proportion of major and minor fatty acids (Zheng et al., 2023; Osorio et al., 2023).

Furthermore, the stage of seed maturation directly influences the chemical profile of the oil, since lipid accumulation occurs progressively throughout fruit development, accompanied by changes in the composition of secondary metabolites and other compounds associated with the lipid fraction (Tidke et al., 2024). Post-harvest conditions, such as storage and exposure to light, oxygen, and moisture, may also promote lipid oxidation processes and degradation of specific constituents, thereby altering the observed chemical profile (Zhang F. et al., 2021; Zhang R. et al., 2021). Thus, the heterogeneity observed in the oil composition among the samples may be interpreted as the result of interactions among environmental, physiological, and technological factors, which affect both the biosynthesis and stability of the metabolites present in the lipid fraction (Cao et al., 2025).

Regarding the molecular networks associated with the physicochemical parameters, the higher intensity of certain compounds within elevated or intermediate ranges, as observed in the refractive index and density networks, is directly related to their structure and chemical state. In other words, compounds showing greater intensity within these ranges correspond to long-chain lipids, higher molecular mass compounds, and molecules with a greater degree of unsaturation, such as fatty acids and their esters. These compounds exhibit higher electronic polarizability and stronger intermolecular interactions, which simultaneously increase both refractive index and density (Zidane et al., 2026).

In contrast, the network associated with the acidity index demonstrated greater variability among the ranges, with the presence of different acidity levels within the same cluster (Vasconcelos Santos et al., 2024). The observed results indicate that structurally related compounds may be associated with distinct levels of lipid degradation, reflecting the influence of processes such as hydrolysis and oxidation. In other words, the basic composition of the oil may remain relatively similar, while the degree of lipid degradation varies more markedly among samples, exerting a stronger effect on this parameter than on n physical properties such as density and refractive index (Mehany et al., 2026; Vasconcelos Santos et al., 2024).

Principal Component Analysis (PCA)

PCA evaluation was performed through the simultaneous analysis of the score and loading plots. Analysis of the score plot (Supplementary Figure S1A) revealed a discrete clustering pattern involving samples 1, 3, 4, 6, 8, 9, and 10 along the positive axis of PC1. Examination of the loading plot (Supplementary Figure S1B) showed that the variables were positioned along the negative axis of PC1, suggesting that the samples forming this cluster present lower concentrations of fatty acids. It was also observed that samples 1 and 9 were slightly separated within the group due to their higher density and refractive index values compared to the remaining samples. Samples 2, 5, and 7 did not cluster with the other samples because they exhibited slightly higher fatty acid concentrations than those observed in the clustered group. Aytar and Aydın 50 also reported high concentrations of octadecanoic and eicosanoic acids in green coffee samples (Aytar and Aydın, 2025).

Additionally, analysis of the loading plot showed that fatty acids such as palmitic, linoleic, stearic, and eicosanoic acids presented similar vector orientations, indicating a positive correlation among these compounds and suggesting a joint contribution to the chemical variability of the oil. In contrast, physicochemical parameters such as density and refractive index exhibited distinct orientations within the factorial space, indicating that these properties contribute independently to sample discrimination (Correa et al., 2024; Ojha et al., 2024).

The relationship observed between fatty acids and physicochemical parameters directly reflects the influence of lipid composition on the structural properties of the oil. Saturated and monounsaturated fatty acids, such as palmitic and oleic acids, exhibit greater structural stability, which tends to influence physical parameters such as density and refractive index. In contrast, unsaturated fatty acids, such as linoleic acid, present greater chemical reactivity and higher susceptibility to hydrolysis and oxidation, processes that contribute to an increase in the acidity index through the formation of free fatty acids (García-Ramón et al., 2025). Thus, variations in the relative abundance of these compounds directly reflect changes in the physicochemical parameters of the oil, demonstrating that differences in lipid composition are determinant factors for the physicochemical properties observed among the analyzed samples (García-Ramón et al., 2025).

The present study enabled the physicochemical and metabolomic characterization of oil extracted from green seeds of C. arabica L. cultivated in the southwestern region of Bahia, Brazil. The results demonstrated that the samples exhibited satisfactory values for density and refractive index, whereas the acidity index showed greater variability among the analyzed oils. Metabolomic analysis by GC–MS combined with molecular networking enabled the identification of fatty acids and other metabolites associated with the lipid fraction, revealing differences in the chemical profiles among the samples. Additionally, multivariate analyses demonstrated an association between chemical composition and the evaluated physicochemical parameters, contributing to the understanding of the variability observed among the oils. Overall, the obtained results demonstrate that the integrated characterization of C. arabica L. oil produced in the southwestern region of Bahia provides robust chemical and metabolomic foundations for understanding the quality of this natural product.

Statements

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Funding

The author(s) declared that financial support was received for this work and/or its publication. The authors gratefully acknowledge financial support from the National Council for Scientific and Technological Development (CNPq, Conselho Nacional de Desenvolvimento Científico e Tecnológico, grant # 408549/2024-6), the State of Sao Paulo Research Foundation (FAPESP, Fundação de Amparo à Pesquisa do Estado de São Paulo, grant # 2025/27006-6), and from Coordination for the Improvement of Higher Education Personnel (CAPES–Finance Code 001).

Data availability

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

Author contributions

JO: Writing – original draft, Formal Analysis, Investigation. PL: Formal Analysis, Investigation, Writing – original draft. MS: Formal Analysis, Writing – original draft, Methodology. MC: Writing – original draft, Formal Analysis, Methodology. LB: Formal Analysis, Methodology, Writing – original draft. ID: Writing – original draft, Formal Analysis, Methodology. BN: Writing – original draft, Data curation. AS: Supervision, Writing – original draft, Formal Analysis, Methodology. GD: Supervision, Writing – original draft, Writing – review and editing. JA: Conceptualization, Funding acquisition, Project administration, Supervision, Writing – original draft, Writing – review and editing.

Acknowledgements

The authors thank Izabel Cristina Casanova Turatti, Dr. Jacqueline Nakau Mendonça, Prof. Dr. Norberto Peporine Lopes, and the Center for Research in Natural and Synthetic Products (NPPNS) for the analysis of compounds by mass spectrometry, as well as the financial support granted by the National Council for Scientific and Technological Development (CNPq), the PIBIC Program for two undergraduate research scholarships, and CAPES (process no. 88882.461677/2019-01).

References

  1. Agência Nacional de Vigilância Sanitária (ANVISA) (2023). * Farmacopeia Brasileira. 7a ed. Brasília: ANVISA.
  2. Amiri R., Akbari M., Moradikor N. (2024). Bioactive potential and chemical compounds of coffee. Prog. Brain Res. 23 – 33. doi:10.1016/bs.pbr.2024.06.011
  3. Aytar E. C., Aydın B. (2025). RETRACTED ARTICLE: investigation of chemical composition, antioxidant properties and molecular docking in different roasting stages of coffee beans. Food Bioprocess Technol. 18, 1464 – 1482. doi:10.1007/s11947-024-03539-1
  4. Baldelli A., Aguilera J. M. (2025). Size reduction and particle size influence the quantification of phenolic-type compounds and antioxidant activity in plant food matrices. Trends Food Sci. and Technol. 162, 105081. doi:10.1016/j.tifs.2025.105081
  5. BARRA DO CHOÇA (2023). Café de Barra do Choça ganha o mundo e está nas redes McDonald's e Starbucks dos Estados Unidos e Europa. Available online at:.
  6. Barth T., Aytar E. C., Aydın B., Ferreira L. D. S., Lopes N. P., Moraes de Oliveira A. R. (2015). In vitro metabolism of the lignan (−)-grandisin, an anticancer drug candidate, by human liver microsomes. Drug Test. Anal. 7 (9), 780 – 786. doi:10.1002/dta.1779
  7. Brondani Teixeira Ribas F., Gasparetto H., Gonçalves Salau N. P. (2025a). The hidden dimension of oil extraction using green solvents on the physicochemical and structural properties of crude oil and defatted meals. Food Chem. 494, 146201. doi:10.1016/j.foodchem.2025.146201
  8. Brondani Teixeira Ribas F., Gasparetto H., Gonçalves Salau N. P. (2025b). Mass transfer study in soybean oil extraction using ethanol and mixtures with p-cymene as co-solvent: early uses of pore volume and surface diffusion model. J. Mol. Liq. 428, 127571. doi:10.1016/j.molliq.2025.127571
  9. Cao M., Zhang B., Duan M., Zhang H., Yan S., Yang F. (2025). Dynamic changes in fatty acids in macadamia fruit during growth and development. Agronomy 15 (7), 1682. doi:10.3390/agronomy15071682
  10. Constantini J. V. C., Drudi Lepore E., Sorrechia R., Cristina Baccetti Medeiros C., Veloso de Paula A., Pietro R. C. L. R. (2023). Caracterização físico-química do óleo de café verde (coffea arabica l.) obtido na cidade de guaxupé – MG. Rev. Multidiscip. Educ. Meio Ambiente 4 (3). doi:10.51189/integrar/rema/3692
  11. Correa K. L., Brito F., Mourão E. S., Caroline H., Christtine S., Louze M. (2024). Physicochemical and nutritional properties of vegetable oils from Brazil diversity and their applications in the food industry. Foods 13 (10), 1565. doi:10.3390/foods13101565
  12. Dantas R. C. (2017). “ Caracterização físico-química do óleo de café verde,” Bachelor’s Thesis. Natal, Brazil: Institute of Chemistry, Federal University of Rio Grande do Norte.
  13. dos Santos D. G., Coelho C. C. de S., Ferreira A. B. R., Freitas-Silva O. (2021). Brazilian coffee production and the future microbiome and mycotoxin profile considering the climate change scenario. Microorganisms 9 (4), 858. doi:10.3390/microorganisms9040858
  14. EMBRAPA. Produção De Café Arábica Corresponde a 64% E Café Conilon a 36% Da Safra Total Dos Cafés Do Brasil Em 2022 (2022). Available online at:.
  15. Galarza G., Figueroa J. G. (2022). Volatile compound characterization of coffee (Coffea arabica) processed at different fermentation times using SPME–GC–MS. Molecules 27 (6), 2004. doi:10.3390/molecules27062004
  16. García-Ramón D. F., Cios J., Soares M. E. M., Sant'Ana J. P. F., Viganó J., Rodrigues C. E. C. (2025). Sustainable aqueous extraction of Brazilian varieties avocado oil: effects on yield, physical and chemical properties, fatty acids and triacylglycerols. Food Chem. 493 (Pt 1), 145693. doi:10.1016/j.foodchem.2025.145693
  17. Gebreegziabher B. W., Adaramola M. S., Morken J., Dubale A. A. (2025). Optimization of solid-liquid extraction: solvent effects on the fatty acid composition and physicochemical properties of Argemone mexicana seed oil and valorization of seed residue. Industrial Crops Prod. 233, 121468. doi:10.1016/j.indcrop.2025.121468
  18. Genovese A., Caporaso N., Baiano A. (2025). The impact of brewing methods on the quality of a cup of coffee. Beverages 11 (5), 125. doi:10.3390/beverages11050125
  19. Gharby S., Asbbane A., Ahmed M. N., Gagour J., Hallouch O., Oubannin S. (2025). Vegetable oil oxidation: mechanisms, impacts on quality, and approaches to enhance shelf life. Food Chem. X 28, 102541. doi:10.1016/j.fochx.2025.102541
  20. Gu Y., Shi L., Wu J., Hu S., Shang Y., Hassan M. (2024). Quantitative prediction of acid value of camellia seed oil based on hyperspectral imaging technology fusing spectral and image features. Foods 13 (20), 3249. doi:10.3390/foods13203249
  21. Guia do Café (2024). Práticas Comerciais Relevantes Para Exportadores Em Países Produtores De Café.
  22. Herculano L. S., Lukasievicz G. V. B., Sehn E., Torquato A. S., Belançon M. P., Savi E. (2021). The correlation of physicochemical properties of edible vegetable oils by chemometric analysis of spectroscopic data. Spectrochimica Acta Part A Mol. Biomol. Spectrosc. 245, 118877. doi:10.1016/j.saa.2020.118877
  23. Khaled B. M., Biswas M., Sina A. A., Alam A., Alim M. A., Hossain M. E. (2024). Detection of the changes in fatty acid composition and physicochemical properties in selected edible oils heated at different degrees of time. Food Chem. Adv. 4, 100716. doi:10.1016/j.focha.2024.100716
  24. Lima P. M., Guimarães C. S., Gonçalves C. V., Pereira C. C., Garcia L. B., Santos I. D. (2025). Synergistic antimicrobial effects of coffee green extract and oil microemulsions enhanced by photodynamic therapy: a strategy against antibiotic resistance. J. Drug Deliv. Sci. Technol. 110, 107094. doi:10.1016/j.jddst.2025.107094
  25. Lin Y., Wang Y., Li Y. (2025). Exploring alternative solvents to n-hexane for green extraction of lipid from camellia oil cakes. Food Chem. X 27, 102443. doi:10.1016/j.fochx.2025.102443
  26. Lutz I. A. (2008). Normas Analíticas do Instituto Adolfo Lutz: métodos físicos e químicos para análises de alimentos. 4.ed.
  27. Makiso Urugo M., Tola Y. B., Ogah O., Liben Endale F. (2023). Bioactive compounds in coffee and their role in lowering the risk of major public health consequences: a review. Food Sci. Nutr. 12 (2), 734 – 764. doi:10.1002/fsn3.3848
  28. Maressa P. (2013). Óleo De Grãos De Café Verde Obtidos Por Extração Supercrítica, Extração Com Fluído Pressurizado E Comparação De Sua Composição Com O Óleo De Café Extraído Pelos Pequenos Produtores Do Estado De São Paulo: Análise Da Concentração De Compostos Bioativos.
  29. Mehany T., González-Sáiz J. M., Pizarro C. (2026). Rapid monitoring and quantification of primary and secondary oxidative markers in edible oils during deep frying using near-infrared spectroscopy and chemometrics. Foods 15 (3), 557. doi:10.3390/foods15030557
  30. Muñoz A. E., Hernández S. S., Tolosa A. R., Burillo S. P., Olalla Herrera M. (2020). Evaluation of differences in the antioxidant capacity and phenolic compounds of green and roasted coffee and their relationship with sensory properties. LWT 128, 109457. doi:10.1016/j.lwt.2020.109457
  31. Ojha P. K., Poudel D. K., Rokaya A., Maharjan S., Timsina S., Poudel A. (2024). Chemical compositions and essential fatty acid analysis of selected vegetable oils and fats. Compounds 4 (1), 37 – 70. doi:10.3390/compounds4010003
  32. Osorio P. V., Matallana Pérez L. G., Fernandez-Alduenda M. R., Alvarez Barreto C. I., Gallego Agudelo C. P., Montoya Restrepo E. C. (2023). Chemical composition and sensory quality of coffee fruits at different stages of maturity. Agronomy 13 (2), 341. doi:10.3390/agronomy13020341
  33. Pereira E. B. (2016). Potencial dos óleos de café verde e de soja na produção de biodiesel via rota etílica. Rev. Univ. Vale Rio Verde. 14 (1), 464473.
  34. Pereira W. O. (2018). Agronegócio Do Café Em Vitória Da Conquista-BA: Agricultura Capitalista Mundializada E Reprodução Fictícia De Capital.
  35. Piasek A. M., Bardadyn P., Trojan Z., Jelonek K., Wysocki Ł., Kobiela T. (2026). Research on spent coffee grounds: from oil extraction to its potential application in cosmetics. Waste Biomass Valor. 17 (6), 33393355. doi:10.1007/s12649-025-03284-2
  36. Popova M., Bankova V. (2023). Contemporary methods for the extraction and isolation of natural products. BMC Chemistry 17 (1), 68. doi:10.1186/s13065-023-00960-z
  37. Ramos P. R., Sponchiado J., Febrônio V., Dacanal G. C., Lopes A. (2024). Kinetics of vegetable oils (Rice bran, sunflower seed, and soybean) extracted by pressurized liquid extraction in intermittent process. Processes 12 (6), 1107. doi:10.3390/pr12061107
  38. Ruse G., Jîjie A.-R., Moacă E.-A., Pătraşcu D., Ardelean F., Jojic A.-A. (2025). Coffea arabica: an emerging active ingredient in dermato-cosmetic applications. Pharmaceuticals 18 (2), 171. doi:10.3390/ph18020171
  39. Schoss K., Glavač N. K. (2024). Supercritical CO2 extraction vs. hexane extraction and cold pressing: comparative analysis of seed oils from six plant species. Plants 13 (23), 3409. doi:10.3390/plants13233409
  40. Thapa D., Kumar V., Naik B., KumarGupta A., Richa R. (2025). Process optimization for the extraction of Shorea robusta (Sal) seed oil using response surface methodology and its characterization: a sustainable approach to oil valorization. Food Chem. X 32, 103380. doi:10.1016/j.fochx.2025.103380
  41. Tidke A., Das R., Annapure U. S., Kundu D. (2024). Solvent optimization for caffeine and tannic acid extraction from guarana seed: insights from molecular dynamics simulations. J. Mol. Liq. 416, 126462. doi:10.1016/j.molliq.2024.126462
  42. Tsukui A., Oigman S. S., Rezende C. M. (2014). Óleo de Grãos de Café Cru: Diterpenos Cafestol e Caveol. Rev. Virt. Quím. 6, 13 – 33. doi:10.5935/1984-6835.20140003
  43. Vasconcelos Santos O., Silva Lemos Y., Viera R., Teixeira-Costa B. E. (2024). Lipids from the purple and white açaí (Euterpe oleracea Mart) varieties: nutritional, functional, and physicochemical properties. Front. Nutr. 17, 11. doi:10.3389/fnut.2024.1385877
  44. Wang M., Carver J. J., Phelan V. V., Sanchez L. M., Garg N., Peng Y. (2016a). Compartilhamento e curadoria comunitária de dados de espectrometria de massa com a Global Natural Products Social Molecular Networking. Nat. Biotechnology 34 (8), 828.
  45. Wang M., Carver J. J., Phelan V. V., Sanchez L. M., Garg N., Peng Y. (2016b). Sharing and community curation of mass spectrometry data with global natural products social molecular networking. Nat. Biotechnol. 34 (8), 828 – 837. doi:10.1038/nbt.3597
  46. Wang Y., Wang X., Du P., Liu X., He S., Li L. (2025). Lipidomic profiling provides insights on arabica coffee flavor diversity in different postharvest processing methods. Curr. Res. Food Sci. 10, 101108. doi:10.1016/j.crfs.2025.101108
  47. Zhang F., Li Z., Zhou J., Gu Y., Tan X. (2021a). Comparative study on fruit development and oil synthesis in two cultivars of Camellia oleifera. BMC Plant Biol. 21 (1), 348. doi:10.1186/s12870-021-03114-2
  48. Zhang R., Zhu Z., Jia W. (2021b). Time-Series lipidomics insights into the progressive characteristics of lipid constituents of fresh walnut during postharvest storage. J. Agric. Food Chem. 69 (46), 13796 – 13809. doi:10.1021/acs.jafc.1c05120
  49. Zheng T., Zeng Ht., Sun By. (2023). Avaliações multiambientais em diferentes regiões ecológicas revelam os efeitos do clima e do solo no conteúdo de amidas em cascas de Zanthoxylum bungeanum Maxim. (espinhos-de-cabra-chinesa). BMC Plant Biol. 23, 313. doi:10.1186/s12870-023-04328-2
  50. Zhu J., Zhou L., Zhao M., Wei F., Fu H., Marchioni E. (2023). Revealing the dynamic changes of lipids in coffee beans during roasting based on UHPLC-QE-HR-AM/MS/MS. Food Res. Int. 174, 113507. doi:10.1016/j.foodres.2023.113507
  51. Zidane O. D., Telli A., Manaa A., Goussi R., Jouira H. B., Khaled H. (2026). Lipid composition and functional properties of quinoa (Amarilla Sacaca) seed oil: impact of cultivation in a hyper-arid Saharan environment on yield, fatty acids, and antioxidant capacity. Front. Sustain. Food Syst. 10, 10. doi:10.3389/fsufs.2026.1765270

Citation

Oliveira, J. F. d. S., et al. (2026). Physicochemical characterization and metabolomic analysis of green coffee oil from Coffea arabica L. cultivated in Bahia, Brazil. Frontiers in Natural Products, Volume 5. https://doi.org/
Received
Accepted
Published
Volume
5 - 2026

Copyright

Copyright © 2026 Oliveira, Lima, Sousa Marinho, Coutinho Silva, Beatriz de Andrade Silva, Dias dos Santos, Neves, Souza, Damasceno and Amaral.

License: https://creativecommons.org/licenses/by/4.0/

*Correspondence: Juliano Geraldo Amaral

Article metrics

Views
0
Downloads
0
Citations
0
Shares
0

Related articles

Original ResearchFlavonoid profiling and antioxidant activity of Passiflora setacea DC. leavesAndré Santos Alkimim, Maria Luiza Nitzsche Morato, Caroline Roberta Leal da Mota, Marina Pereira Rocha, Juliana Rodrigues Vasconcelos, Claudinei da Silva Souza, Júlio César Dias Lopes, Wagner Campos Otoni, Cláudia Gontijo Silva, Vera Lucia Almeida, Priscilla Rodrigues Valadares CampanaOriginal ResearchIsolation and identification of a new biflavonoid from the Amazonian tree Ceiba pentandra (L.) Gaertn. (sumaúma)Priscila Brasil Augusto de Souza, Davi Santos Oliveira, Estefane de Oliveira Picanço, Magno Perêa Muniz, Claudete Catanhede do Nascimento, Maria da Paz LimaOriginal ResearchMicroemulsion-based delivery of peppermint essential oil as a sustainable strategy for Aedes aegypti controlCarolina Paula de Souza Moreira, Natasha Reynaud, Marcos Paulo Gomes Mol, Gilson Faria, Sérgio Caldas