Impact of roasting levels and brewing cycles on bioactive compounds in spent coffee grounds

Abstract

Spent coffee grounds (SCGs) represent a potential source of residual bioactive compounds for sustainable reuse. Effects of roasting levels and sequential brewing cycles on recovery of total phenolic content (TPC), total flavonoid content (TFC), caffeine, and chlorogenic acid (CGA) from Arabica and Robusta SCGs were investigated. Coffee beans were roasted (light, medium, dark), brewed through three hydrothermal cycles, and the resulting SCGs extracted with 70% ethanol. The first brewing cycle removed most water-soluble bioactive compounds, while subsequent brews induced smaller compositional changes, indicating the persistence of functional compounds. Roasting influenced the initial bioactive profile, but its impact diminished after brewing. Robusta SCGs retained higher TPC and antioxidant activity while caffeine diminished with brewing cycle, they were relatively stable to roasting while CGA was more heat-sensitive. Principal component analysis confirmed brewing history as the main factor governing SCG chemical profiles. These findings support brewing-informed SCG valorization for sustainable functional food applications.

Highlights

• •Light roasting preserves phenolics, flavonoids, CGA, and antioxidant activity.
• •Brewing cycles reduce soluble bioactives while increasing SCG moisture/porosity.
• •Caffeine remains stable; chlorogenic acid shows strong heat sensitivity.
• •Robusta retains slightly more bioactive compounds than Arabica overall.
• •PCA confirms roasting level and brewing cycle as key drivers of composition.

1. Introduction

Bioactive compounds derived from food by-products have become an area of growing interest in food science and sustainability research. Among the wide range of agro-industrial wastes, spent coffee grounds (SCGs) are particularly noteworthy due to the vast quantities generated globally and their high content of valuable compounds. Coffee is one of the most consumed beverages and most traded commodities worldwide, with Coffea arabica (Arabica) and Coffea canephora (Robusta) being the two main species cultivated and consumed (Mihai et al., 2024). During brewing, a variety of soluble components, including flavors, pigments, and bioactive molecules, are extracted into the beverage. The composition of these compounds is influenced by coffee variety and roasting level, while extractability depends on factors such as grind size and solvent characteristics. The remaining insoluble residue, SCGs, are typically discarded. As global coffee consumption continues to increase, SCG generation has risen accordingly (Cho et al., 2025). With annual global coffee production of approximately 10 million tonnes, an estimated six million tonnes of SCGs are generated worldwide each year, posing significant environmental challenges if not properly managed (Brzezińska et al., 2023). However, these residues still contain significant nutrients and bioactive compounds, positioning them as a promising resource for value-added applications.

SCGs are rich in carbohydrates, lipids and proteins; phenolic acids such as chlorogenic, caffeic, and ferulic acids; flavonoids and other polyphenolics; purine compounds such as caffeine and xanthines; melanoidins; diterpenes; carotenoids; and other antioxidants (Saxena et al., 2025). These compounds have attracted considerable attention for their biological activities, particularly their antioxidant properties, which help mitigate oxidative stress and the risk of chronic diseases (Muscolo et al., 2024). Consequently, the valorization of SCGs has been proposed for the development of functional food ingredients, nutraceuticals, pharmaceuticals, and bio-based materials.

The efficiency of recovering bioactive compounds from SCGs depends on several factors, including the extraction method, solvent type, coffee variety, roasting degree, and brewing cycle (Yildirim et al., 2023). Among solvent systems, ethanol–water mixtures are recognized as particularly effective, offering a favorable balance between polarity and extraction efficiency (Maiyah et al., 2025). Such mixtures enable the recovery of both hydrophilic and moderately hydrophobic compounds, while remaining suitable for industrial applications in food and health products.

In our previous study, we investigated the recovery of bioactive compounds and antioxidants from medium-roasted SCGs using hydrothermal brewing cycles with water and ethanol as solvents. That work demonstrated the importance of the number of brewing cycles in determining compound availability and showed that ethanol extraction generally yielded higher concentrations of bioactive compounds compared to water (Maiyah et al., 2025).

Building on those findings, we now extend the investigation to examine how roasting influences this process. Roasting is known to alter coffee chemistry in that, lighter roasts tend to preserve more chlorogenic acids, while darker roasts favor the formation of Maillard-derived compounds (Freitas et al., 2024). To explore these dynamics, the present study applies ethanol-based extraction to SCGs prepared from both Arabica and Robusta coffees across multiple roasting degrees and sequential brewing cycles.

Therefore, the objective of this study was to evaluate the recovery of bioactive compounds, including total phenolic content (TPC), total flavonoid content (TFC), caffeine, and chlorogenic acid (CGA), from Arabica and Robusta SCGs extracted with ethanol, across three roasting levels (light, medium, and dark) and multiple brewing cycles. By integrating these factors, this work provides a more comprehensive understanding of SCGs as a source of functional compounds that have industrial applications, while simultaneously promoting sustainable waste management processes.

2. Materials and methods

2.1. Materials

Green coffee beans of two varieties, Arabica (Coffea arabica L.) and Robusta (Coffea canephora robusta), were sourced from local farmers in Chiang Rai Province, Thailand. The beans were processed immediately after harvest using the washed method and dried to reach moisture contents of 9.85 ± 0.27% (Arabica) and 10.37 ± 0.14% (Robusta). The dried green beans were stored at ambient temperature in a dry and dark environment prior to roasting. After roasting and grinding, the coffee was packaged in resealable aluminum foil laminated pouches (250 g) provided by Akha Mino Coffee (Chiang Rai, Thailand) and stored at ambient temperature under dry and dark conditions prior to analysis.

Analytical-grade reagents and standards were employed in all experiments. Ethanol was purchased from Fisher Scientific (Leicestershire, UK) and used as the extraction solvent. Gallic acid, quercetin, caffeine, and chlorogenic acid standards were obtained from Sigma Aldrich (USA) and Thermo Scientific (Pittsburgh, USA). Folin–Ciocalteu reagent, aluminum chloride, sodium carbonate, sodium hydroxide, and sodium nitrite were of analytical grade and sourced from standard suppliers. All chemicals were used without further purification.

2.2. Sample preparation

The Arabica and Robusta beans were roasted using a Toper roasting machine (Model TKMSX5, Toper, Turkey) at three levels: light (195 °C, 720 ± 10 s), medium (205 °C, 1200 ± 10 s), and dark (220 °C, 1500 ± 10 s), following the method of Kim et al. (2022) with modification. The samples were designated as Arabica light- (AL), Arabica medium- (AM), Arabica dark- (AD), Robusta light- (RL), Robusta medium- (RM), and Robusta dark- (RD) roasted coffee, respectively. SCGs were prepared using sequential brewing cycles (one, two, or three cycles). Specifically, 7 g of ground coffee was brewed at 92–95 °C with 70 mL distilled water for 20 s using a coffee maker (Model RESESPSK1203, Cuizimate, Thailand). The resulting SCGs were collected after each cycle (SCG1, SCG2, SCG3).

Moisture content (MC) of fresh SCGs was determined by Official Method 934.01 (AOAC, 2002). The samples were dried at 60 °C for 24 h to achieve <10% MC, ground and sieved to a particle size of ∼450 μm using a hand siever, then stored at −20 °C until extraction.

2.3. Extraction of bioactive compounds

Dried SCGs were extracted using 70% ethanol (v/v) based on the method of (Angeloni et al., 2021). A 1:10 solid-to-solvent ratio (w/v) of SCG powder to solvent was used. Samples were sonicated at 40 kHz for 1 h at 25 °C, then centrifuged at 10,000 ×g for 10 min (Model 5920R, Eppendorf, USA). The supernatant was filtered through Whatman No. 4 filter paper and stored at −20 °C until analysis. Each extraction was performed in triplicate.

2.4. Analysis and characterization

2.4.1. Moisture content, color, and browning index

Moisture content was determined using Official Method 934.01 (AOAC, 2002). Wet SCGs (∼2 g) were dried in a hot air oven at 105 °C for 24 h until constant mass was achieved, then MC (%) was calculated on a wet basis from the initial and final mass.

The color of the SCGs was measured using Chroma Meter CR-400 (Konica Minolta Co., Ltd., Japan) and the D65 illuminant system. The measurements were performed in triplicate for each sample, and mean values were calculated for each color parameter. The instrument was initially calibrated with a white standard plate. Color values were expressed as L^⁎, a^⁎ and b^⁎, with triplicate readings. The total color difference (ΔE^⁎) and browning index (BI) were calculated as:

$${\mathrm{\Delta E}}^{\ast }=\sqrt{{\left(L_1^{\ast }-L_0^{\ast }\right)}^2+{\left(a_1^{\ast }-a_0^{\ast }\right)}^2+{\left(b_1^{\ast }-b_0^{\ast }\right)}^2}$$ (1)

$$\mathit{BI}=\frac{100\left(X-0.31\right)}{0.17}$$ (2)

where, X was calculated from the equation:

$$X=\frac{a^{\ast }+1.75L^{\ast }}{5.645L^{\ast }+a^{\ast }-0.30121b^{\ast }}$$ (3)

and $L_1^{\ast }$, $a_1^{\ast }$and $b_1^{\ast }$are the color values of the samples, while $L_0^{\ast }$, $a_0^{\ast }$ and $b_0^{\ast }$ are color values of the white plate (Kieu Tran et al., 2020).

2.4.2. Microstructure analysis by FE-SEM

The surface morphology and microstructural characteristics of the dried SCGs were examined using field emission scanning electron microscopy (FE-SEM). Prior to imaging, samples were sputter-coated with a thin layer of gold–palladium (Au:Pd, 60:40) using a Quorum Q150R ES Plus coater (Quorum Technologies, East Sussex, UK) at room temperature (∼25 °C) to prevent charging under the electron beam. The coated samples were placed in an Apreo 2 FE-SEM system (Thermo Fisher Scientific, Carlsbad, CA, USA) and analyzed under high vacuum conditions at an accelerating voltage of 15 kV. Imaging was performed in triplicate for each sample.

2.4.3. Fourier transform infrared (FTIR) spectroscopy

FTIR analysis was used to identify the major functional groups in coffee samples and their SCGs after successive brewing cycles (SCG1–SCG3). The chemical analysis was performed on dried SCGs using an Invenio FTIR spectrometer (Bruker, Germany). Spectra were recorded in the range of 4000–400 cm⁻¹ using the attenuated total reflectance (ATR) mode with a spectral resolution of 4 cm⁻¹ and 64 scans per sample. A small amount of ground sample was placed directly on the ATR crystal without additional preparation. This technique was selected due to its suitability for analyzing small sample sizes and for detecting surface-level chemical changes. Each analysis was performed in triplicate.

2.4.4. Total phenolic content

Total phenolic content (TPC) was determined using the Folin–Ciocalteu European Commission Regulation method modified by Bobo-García et al. (2015). Extract (20 μL) was mixed with 100 μL Folin–Ciocalteu reagent (25%), followed by 75 μL sodium carbonate (10%). After incubation (2 h, ∼25 °C), absorbance was measured at 750 nm (Ensight™, PerkinElmer, USA). Results were expressed as mg gallic acid equivalents (GAE)/g dry SCG.

2.4.5. Total flavonoid content

Total flavonoid content (TFC) was quantified following Bijla et al. (2022) with minor modifications. An aliquot of extract (100 μL) was mixed sequentially with 60 μL of 5% sodium nitrite, 50 μL of 10% aluminum chloride, and 30 μL of 0.1 M sodium hydroxide, allowing the mixture to stand for 5 min after each addition. Absorbance was then measured at 510 nm using a Shimadzu UV-1800 spectrophotometer (Japan). Results were expressed as mg quercetin equivalents (QE) /g dry SCG.

2.4.6. Caffeine and chlorogenic acid content

Caffeine and chlorogenic acid (CGA) were quantified following Alamri et al. (2022) with modifications. Extracts (0.5 mg/mL) were partitioned with dichloromethane to isolate caffeine, and absorbance was measured at 274 nm using a UV–Vis spectrophotometer (Shimadzu UV-1800, Japan). CGA was quantified from the residual aqueous phase at 324 nm. Calibration curves were prepared using caffeine and CGA standards, respectively.

2.4.7. Antioxidant activity by ABTS

The ABTS radical scavenging capacity of the SCG extracts was determined following the method of Brzezińska et al. (2023) with minor modifications. The ABTS radical cation (ABTS•⁺) was prepared by reacting 7.5 mM ABTS solution with 2.4 mM potassium persulfate (1:1, v/v) and allowing the mixture to stand in the dark at room temperature (∼25 °C) for 12 h. Prior to use, the ABTS•⁺ solution was diluted with 70% ethanol to obtain an absorbance of 0.70 ± 0.02 at 734 nm.

For the assay, 10 μL of each SCG extract was mixed with 290 μL of the diluted ABTS•⁺ solution in a 96-well plate. After incubation in the dark for 6 min, absorbance was recorded at 734 nm using a microplate reader (Ensight™, Multimode, PerkinElmer, USA). A standard curve was prepared using Trolox (0–500 μg/mL in ethanol), and results were expressed as mg Trolox equivalents per g of dry SCG (mg TE/g DW). All analyses were performed in triplicate.

2.4.8. Antioxidant activity by FRAP

The ferric reducing antioxidant power (FRAP) of the SCG extracts was evaluated according to Mussatto et al. (2011) with slight modifications. The FRAP reagent was freshly prepared by mixing 300 mM sodium acetate buffer (pH 3.6), 10 mM TPTZ solution (dissolved in 40 mM HCl), and 20 mM FeCl_₃·6H_₂O solution in a ratio of 10:1:1 (v/v/v).

For the assay, 10 μL of each extract was combined with 290 μL of FRAP reagent in a 96-well microplate and incubated at 37 °C for 15 min. Absorbance was then measured at 593 nm against a distilled water blank. A calibration curve was constructed using Trolox standard solutions (0–500 μg/mL in ethanol). The results were expressed as mg Trolox equivalents per g of dry SCG (mg TE/g DW). Each measurement was conducted in triplicate.

2.5. Statistical analysis

All experiments were conducted in triplicate and results were expressed as mean ± standard deviation (SD). One-way analysis of variance (ANOVA) followed by Tukey's test was performed to evaluate significant differences among samples at P < 0.05. Principal component analysis (PCA) was applied to explore the relationship between bioactive compounds (TPC, TFC, caffeine, CGA) and antioxidant activities (ABTS, FRAP) across coffee and SCGs. PCA was carried out using OriginPro 2024 (OriginLab Corporation, Northampton, MA, USA) based on correlation matrices, and results were visualized as score and loading plots.

3. Results and discussion

• 3.1.Moisture content, color, and browning index

3.1.1. Moisture content

Roasting level significantly influenced the initial MC of the coffee grounds (Fig. 1a-b). Light-roasted samples exhibited higher MC (≈5–6%) compared to medium- and dark-roasted samples (≈2–3%). This suggests that most moisture loss occurred during the transition from light to medium roasting, while further roasting did not result in additional measurable moisture reduction. This behavior reflects substantial water loss and structural changes induced by high-temperature roasting, occurring predominantly at the early stages of roasting. These results agree with reports (Maiyah et al., 2025; Tarigan et al., 2022) that more severe roasting promotes extensive water evaporation and structural densification, thereby reducing the bean's capacity to hold residual moisture. SEM micrographs support this observation (Fig. 2), showing that darker roasts possess a more compact and collapsed structure, while lighter roasts maintain a more open and porous surface that allows higher moisture retention.

Fig. 1.

Moisture content of coffee and SCGs of Arabica (a) and Robusta (b) and color attributes of coffee and SCGs of Arabica (c) and Robusta (d). MC was analyzed before dehydration drying, while other parameters were analyzed after dehydration drying. Different superscript letters indicate significant differences (P < 0.05) among coffee grounds and spent coffee grounds from different roasting and brewing levels.

Fig. 2.

SEM analysis of coffee and SCGs of Arabica and Robusta with 1000× magnification.

Successive brewing cycles caused a pronounced rise in MC across all roasting levels, with first-cycle grounds (SCG1) already exceeding 58% and third-cycle samples reaching 65–68%. This increase reflects the leaching of soluble solids, which creates a porous, capillary-rich matrix, as well as the progressive hydration of residual cellulose and hemicellulose remaining on the surface of the grounds (Saxena, 2025). The largest increase occurred between the control and SCG1, consistent with SEM images (Fig. 2) showing initial cell-wall rupture and surface fragmentation, while later cycles produced smaller gains as the structure became saturated with water. These results demonstrate that repeated brewing progressively alters the coffee matrix, enhancing hydration and porosity in line with the observed microstructural changes.

Varietal differences were also evident, as Robusta SCGs tended to exhibit slightly higher MC than Arabica at the same brewing stage (e.g., RD-SCG3 = 68.25% vs. AL-SCG3 = 64.76%). This may be linked to differences in cell wall structure and porosity, which influence water absorption capacity (Freitas et al., 2024).

3.1.2. Color attributes (L^⁎, a^⁎, b^⁎, ΔE^⁎)

As expected, roasting strongly affected the color properties of coffee before brewing (Fig. 1c-d). Light roast Arabica and Robusta (AL-C and RL-C) exhibited the highest L^⁎ values (∼42.5), indicating a lighter appearance, whereas dark roast of Robusta (RD-C) had lower L^⁎ values (∼37.2), reflecting increased pigment formation from Maillard reactions and caramelization. The color differences in lightness and hue with increased roasting are also visually evident in the representative photographs provided in the Supplementary Material (Supplementary Fig. S1). Likewise, light roasted SCGs had higher a^⁎ (redness) and b^⁎ (yellowness) values, both of which decreased for more heavily roasted SCGs.

After brewing, SCGs consistently exhibited lower a^⁎ and b^⁎ values (and thus, lower color saturation) compared to their respective coffee controls, reflecting the leaching of chromophoric compounds such as melanoidins, phenolics, and caramelization products into the beverage. The ΔE^⁎ values revealed perceptible differences between controls and SCGs, with slight increases observed across brewing cycles, indicating gradual changes in color even after the first extraction.

Arabica SCGs had slightly higher a^⁎ and b^⁎ values than Robusta SCGs, suggesting that pigment content and retention is influenced by varietal composition. Arabica is generally richer in chlorogenic acids and related precursors, which, during roasting and brewing undergo degradation and polymerization reactions that form brown-colored melanoidins, contribute to color intensity and influence the reddish-yellow color observed in the SCGs (Silva et al., 2022).

3.1.3. Browning index (BI)

The browning index was highest in light-roasted coffee and lower in medium and dark roasts (Fig. 1c-d). Although darker roasts visually appear more intensely brown due to lower L^⁎ values, the BI calculation suggests that light roasts contain a greater proportion of water-soluble brown pigments. This reflects differences in melanoidin formation, where lighter roasts generate pigments with higher solubility, while darker roasts produce more polymerized, less extractable compounds (Farah, 2019).

Following brewing, the BI of SCGs decreased substantially, typically ranging between 12 and 17. This reduction demonstrates the extraction of soluble brown pigments into the coffee infusion, leaving behind a residue with less chromophoric intensity. Across brewing cycles, BI remained relatively stable, indicating that most soluble pigments were released during the first extraction. Light-roasted SCGs tended to maintain slightly higher BI than medium and dark roasts, supporting the idea that lighter roasting produces more extractable melanoidins (Kulapichitr et al., 2022). The observed BI values are also in line with reports that C. canephora produces significantly higher melanoidins than C. arabica at the same roasting level (Freitas et al., 2024), highlighting intrinsic variety differences in pigment formation and stability.

3.2. Microstructure analysis by FE-SEM

The microstructural characteristics of dried coffee and SCGs were observed using FE-SEM (Fig. 2) to evaluate the effects of brewing cycles and roasting levels on surface morphology. The control samples of both Arabica and Robusta exhibited compact, smooth structures with intact cell walls, indicating minimal disruption prior to extraction. After the first brewing cycle (SCG1), the surfaces became slightly rougher with minor porosity, reflecting the initial leaching of soluble compounds while much of the structural matrix remained preserved. As brewing progressed through the second and third cycles (SCG2 and SCG3), the SCGs showed gradual degradation, with increasingly thin, fragmented, and porous cell walls, indicating extensive extraction of water-soluble compounds such as phenolics, flavonoids, and caffeine. These trends were slightly more pronounced in Robusta, likely due to varietal differences in cell wall composition.

Roasting intensity also influenced SCG morphology. Light-roasted samples maintained more open and porous structures, whereas medium and dark-roasted SCGs appeared more compact and collapsed. This aligns with the understanding that roasting disrupts cell wall structures, particularly cellulose and hemicellulose (Li et al., 2021), thereby increasing polysaccharide solubility and facilitating their extraction during brewing.

3.3. FTIR spectroscopy

The spectra of Arabica (Fig. 3a) and Robusta (Fig. 3b) samples at different roast levels exhibited characteristic absorption bands of roasted coffee, indicating that the fundamental chemical structure is largely preserved even after repeated extraction. A broad band near 3300 cm⁻¹ corresponds to O—H stretching in phenolics, polysaccharides, and residual moisture (Zuluaga et al., 2024), while peaks at 2920–2850 cm⁻¹ reflect aliphatic C—H stretching of lipids (Freitas et al., 2024). Absorptions around 1740–1700 cm⁻¹ are associated with C O stretching from CGA and other phenolic esters (Campbell et al., 2024), and bands at 1600–1500 cm⁻¹ correspond to aromatic C C vibrations of lignin and caffeine. Signals at 1250–1020 cm⁻¹ arise from C—O and C–O–C stretching in polysaccharides such as cellulose and hemicellulose (Thana et al., 2023).

Fig. 3.

FTIR spectra of coffee and SCGs of Arabica (a) and Robusta (b).

Comparison of control coffee grounds with SCG1–SCG3 showed minor decreases in FTIR peak intensities, particularly in the O—H and C O regions, indicating gradual extraction of phenolic and carbohydrate-related compounds across brewing cycles. This reduction aligns with the measured declines in TPC, TFC, and CGA, confirming that repeated extraction depletes the bioactive fraction but does not eliminate it. Importantly, no new absorption peaks were observed, suggesting that brewing removes soluble compounds without generating detectable degradation products. Differences between Arabica and Robusta spectra were subtle and mainly reflected inherent compositional variations. Roasting also influenced the profiles, with darker roasts showing band broadening and slight shifts consistent with phenolic degradation, Maillard reactions, and melanoidin formation. These spectral changes are consistent with the observed reductions in extractable phenolics and flavonoids in dark-roasted SCGs (Moon et al., 2009).

These findings indicate that both Arabica and Robusta SCGs maintain a relatively stable chemical composition throughout successive hydrothermal brewing cycles, suggesting a resilience of their functional groups as suggested by Maiyah et al. (2025).

3.4. Total phenolic content and total flavonoid content

The TPC of SCGs was strongly influenced by both roasting and brewing cycles, reflecting the dynamic changes that phenolic compounds undergo during coffee processing (Fig. 4a,b). Light roasts generally retained higher phenolic levels than medium and dark roasts, highlighting their thermal sensitivity. Prolonged exposure to high temperatures during roasting promotes oxidative degradation and polymerization of phenolics, including their incorporation into melanoidin structures, which reduces the amount of free, extractable compounds (Wu et al., 2022).

Fig. 4.

TPC of coffee and SCGs of Arabica (a) and Robusta (b), and TFC of coffee and SCGs of Arabica (c) and Robusta (d). Different lowercase letters within the same brewing cycle indicate significant differences (P < 0.05) among roasting levels. Different uppercase letters within the same roasting level indicate significant differences (P < 0.05) among brewing cycles.

Brewing cycles exerted a stepwise effect on TPC, reflecting the solubility and accessibility of phenolics in SCGs. The first brew removed most of the water-soluble phenolics, while subsequent cycles yielded progressively lower amounts, consistent with the high solubility of chlorogenic acids. However, measurable TPC remained after three brews, indicating that some phenolics were not fully extracted in the previous cycles due to limited brewing time (Maiyah, 2025). These results highlight that repeated brewing progressively depletes extractable phenolics, but some compounds may require multiple extractions for complete recovery.

In Arabica SCGs (Fig. 4a), TPC of control samples (AL-C ≈ 30, AM-C ≈ 32, AD-C ≈ 30 mg GAE/g) dropped sharply after the first brew (SCG1 ≈ 18–22 mg GAE/g), with further declines in SCG2 (≈13–18 mg GAE/g) and SCG3 (≈10–15 mg GAE/g). A similar trend was observed for TPC in Robusta SCGs (Fig. 4b), with controls (RL-C, RM-C, RD-C ≈ 30–37 mg GAE/g) decreasing to ≈14–22 mg GAE/g by SCG3.

Overall, Robusta SCGs generally exhibited higher TPC values than Arabica, particularly at light and dark roasting levels, which is consistent with the inherently higher phenolic content reported for Robusta beans (Bicho et al., 2013; Maiyah et al., 2025). Notably, at the medium roast level, Arabica samples exhibited comparable or slightly higher TPC values than Robusta, indicating a deviation from the general trend. This suggests that roasting intensity and sequential brewing cycles exerted a stronger influence on TPC behavior than coffee variety under the conditions of this study. In this study, Arabica and Robusta were included as the two most consumed coffee varieties from which bioactives might be recovered. However, varietal differences should be interpreted cautiously, as a rigorous comparison would require stricter control of genetic and agronomic factors beyond the scope of the present work.

Taken together, these findings reveal that TPC in SCGs is governed by a complex interplay of thermal degradation, extraction efficiency, and structural entrapment. Light roasting and initial brewing maximize extractable phenolics, whereas residual compounds in repeatedly brewed SCGs highlight the material's potential as a source of bioactive compounds.

The TPC values were generally higher than those reported by Papageorgiou et al. (2024) for Colombian Arabica SCGs (21.6 ± 0.7 mg GAE/g) and by Mihai et al. (2024) for C. arabica from Cotopaxi (4.19 ± 0.03 mg GAE/g dw) and C. canephora (3.04 ± 0.32 mg GAE/g dw), but lower than those of Zengin et al. (2020), likely reflecting differences in extraction method, sample preparation, and solvent.

The behavior of flavonoids closely mirrored that of phenolics, with roasting level emerging as a major factor influencing their availability in SCGs (Fig. 4c,d). Light roasts consistently retained the highest TFC, whereas dark roasts showed the lowest values. This decline is largely explained by the thermal sensitivity of flavonoids, which undergo degradation, oxidation, and condensation reactions under prolonged heating (Gao et al., 2022). In some cases, medium roasting produced intermediate values, reflecting a balance between compound loss and the release of flavonoids previously bound within the coffee matrix. These results align with earlier studies that demonstrate that increasing roast intensity significantly reduces extractable flavonoids, even though some compounds may persist as part of the Maillard-derived matrix (Maiyah et al., 2025).

Brewing cycles also played a decisive role in flavonoid recovery. As with TPC, the first brew extracted most soluble flavonoids, while the second and third brews yielded progressively smaller amounts. This pattern underscores the high solubility of flavonoids, which are rapidly mobilized during early brewing. Nonetheless, measurable quantities remained in SCGs even after the third cycle, particularly in light and medium roasts. Previous studies support this observation, showing that repeated extractions leave behind a substantial fraction of flavonoids, which can still be recovered through solvent-based methods such as ethanol extraction (Budryn et al., 2012).

Varietal differences further influenced TFC levels, with Robusta SCGs consistently containing higher concentrations than Arabica. This is consistent with Robusta's naturally greater production of secondary metabolites, including flavonoids (Vignoli et al., 2014). Robusta SCGs, particularly from light roasts and earlier brewing cycles, represent a promising raw material for producing flavonoid-rich extracts.

3.5. Caffeine and chlorogenic acid

Caffeine content in SCGs was less affected by roasting than phenolics or flavonoids, reflecting its relative thermal stability (Fig. 5a,b). Light, medium, and dark roasts showed only minor differences in caffeine concentration, though some significant variations were observed across brewing cycles. Robusta consistently contained higher caffeine than Arabica, in line with its naturally higher alkaloid content (Mihai et al., 2024). The caffeine content observed in this study was higher than that reported in some previous studies reviewed by Bicho et al. (2013).

Fig. 5.

Caffeine of coffee and SCGs of Arabica (a) and Robusta (b), and CGA of coffee and SCGs of Arabica (c) and Robusta (d). Different lowercase letters within the same brewing cycle indicate significant differences (P < 0.05) among roasting levels. Different uppercase letters within the same roasting level indicate significant differences (P < 0.05) among brewing cycles.

Brewing cycles had a more pronounced effect on caffeine. The first brew yielded the highest concentrations, followed by gradual declines in the second and third cycles. This pattern reflects caffeine's high solubility in hot water and ethanol, allowing most of the compound to be extracted early, while residual amounts remain trapped within the SCG matrix (Vandeponseele et al., 2021). Notably, measurable caffeine persisted across all roasting levels and brewing cycles, indicating that SCGs remain a viable source of this stimulant for functional food applications or moderate caffeine formulations.

In contrast, CGA levels were influenced by roasting. Light roasts retained significantly higher CGA than medium and dark roasts (Fig. 5c,d), highlighting its thermal sensitivity. Prolonged heating induces degradation and isomerization of CGA, producing lactones and phenylindanes, which contribute to flavor but not the same functional bioactivity (Hasbullah, 2021; Mancini et al., 2018; Pedan et al., 2020). Both Arabica and Robusta exhibited this trend, though Arabica retained slightly more CGA in light roasts.

Brewing cycles rapidly depleted CGA. The first extraction removed much of the compound, with sharp declines in the second and third cycles. This demonstrates that CGA is highly soluble but also more labile than caffeine, leaving only small residual amounts in repeatedly brewed SCGs. Consequently, recovery of CGA is most efficient from lightly roasted SCGs and early brewing cycles.

Varietal differences also determined caffeine and CGA content. Robusta SCGs consistently contained higher levels of caffeine and lower levels of CGA than Arabica, reflecting the natural accumulation of these compounds in Robusta beans (Bicho et al., 2013). The stability of caffeine suggests broader opportunities for recovery, whereas CGA extraction is more limited, favoring lightly roasted Robusta SCGs and early brewing cycles as the optimal source of this bioactive compound.

3.6. Antioxidant activity (ABTS and FRAP)

Fig. 6 shows the antioxidant activity of Arabica (a,c) and Robusta (b,d) SCGs measured by ABTS (a,b) and FRAP (c,d). A clear stepwise decline occurs from the control coffee to the third spent cycle, with the highest activity in the control and first brew followed by marked reductions in subsequent cycles. Light-roasted samples consistently exhibited the greatest antioxidant capacity, medium roasts showed intermediate values, and dark roasts presented the lowest antioxidant activity.

Fig. 6.

ABTS of coffee and SCGs of Arabica (a) and Robusta (b) and FRAP of coffee and SCGs of Arabica (c) and Robusta (d). Different lowercase letters within the same brewing cycle indicate significant differences (P < 0.05) among roasting levels. Different uppercase letters within the same roasting level indicate significant differences (P < 0.05) among brewing cycles.

These patterns indicate that antioxidant activity in SCGs is strongly governed by both roasting degree and brewing cycles. Light-roast SCGs consistently exhibited the highest antioxidant capacity, reflecting the higher retention of phenolic compounds, flavonoids, and CGA. Medium-roast SCGs showed moderate activity, whereas dark-roast SCGs generally had lowest antioxidant activity due to the thermal degradation of these bioactive compounds and other antioxidants during roasting, as widely reported in the literature (Yust et al., 2024).

Brewing cycles produced a clear stepwise decline in antioxidant potential. The first brew consistently provided the highest ABTS radical scavenging and FRAP, followed by significant reductions in the second and third cycles. This can be attributed to the preferential extraction of phenolics, flavonoids, and CGA during the initial brewing, leaving only residual antioxidants in the SCGs. Nevertheless, measurable activity persisted even after the third brew when extracted with ethanol, highlighting the potential of SCGs as a source of functional compounds for food or nutraceutical applications.

Comparing ABTS and FRAP results illustrates complementary aspects of antioxidant behavior. While ABTS reflects radical scavenging activity and FRAP measures reducing power, both assays show similar trends in response to roasting and brewing. The decline in antioxidant activity parallels the reduction in phenolic and flavonoid content, confirming their major contribution. However, the presence of residual activity, particularly in darker roasts, indicates that other compounds, including melanoidins and possibly bound phenolics, also contribute to the overall antioxidant capacity (Patrignani & González-Forte, 2021).

Interestingly, at the medium roasting level, Robusta SCGs exhibited antioxidant activity comparable to or slightly lower than Arabica, a trend also observed for TPC (Fig. 4). This indicates that medium roasting exerts a relatively stronger impact on antioxidant-related compounds in Robusta than in Arabica. One possible explanation is that the higher initial levels of phenolics and caffeine in Robusta may undergo more pronounced thermal degradation or transformation at intermediate roasting conditions, before the formation of antioxidant-active melanoidins becomes dominant at darker roasting stages. Consequently, antioxidant responses to roasting are not strictly proportional to initial phenolic content and may vary between coffee types depending on roasting intensity.

Although some variation between Arabica and Robusta SCGs was observed, this variation was less pronounced than the effects of roasting degree and brewing cycles. Under the conditions of this study, processing factors exerted a stronger influence on antioxidant activity than coffee variety. Therefore, varietal differences should be interpreted cautiously. Nevertheless, similar tendencies have been reported in the literature, with Mihai et al. (2024) also observing higher antioxidant capacity in Robusta compared to Arabica.

3.7. Principal component analysis (PCA)

PCA was applied to investigate the multivariate relationships among bioactive compounds in coffee and SCGs of Arabica and Robusta across different brewing cycles. The PCA score plots (Fig. 7a,b) show a clear separation of samples along the first principal component (PC1), which explained 88.8% and 85.5% of the total variance for Arabica and Robusta, respectively. This indicates that PC1 captures the dominant chemical changes associated with the brewing process.

Fig. 7.

PCA score plots of Arabica and Robusta coffee and SCGs grouped by brewing cycle (a,b) and roasting level (c,d) based on TPC, TFC, caffeine, CGA, and antioxidant activity.

For both coffee varieties, control samples (unbrewed coffee grounds) were distinctly separated from SCG1, SCG2, and SCG3 along PC1, demonstrating that the first brewing cycle removed a substantial proportion of water-soluble bioactive compounds. Samples obtained after subsequent brewing cycles clustered more closely, indicating that additional brewing cycles exerted a progressively smaller effect on the remaining compounds. This pattern is consistent with the univariate results TPC, TFC, caffeine, and CGA, which showed a marked decrease after the first brewing followed by a more gradual decline in later cycles.

Although Arabica and Robusta differ in their initial chemical composition, varietal separation in the PCA score plots was less pronounced than the effect of brewing cycles. The similar clustering trends observed for both varieties indicate that the response of SCGs to repeated brewing follows a comparable pattern regardless of coffee type, suggesting that extraction dominates the multivariate variability of SCGs after brewing.

The PCA loading plots indicate that TPC, TFC, caffeine, CGA, and antioxidant activities (ABTS and FRAP) contributed strongly to PC1, confirming their collective role in differentiating control and brewed samples. PC2 explained a smaller proportion of the variance (5.3% for Arabica and 11.5% for Robusta) and captured minor variations among SCG samples, reflecting differences in the relative abundance of bioactive compounds after repeated extraction rather than changes in compound presence.

The influence of roasting level on PCA separation was comparatively weaker than that of brewing cycles, as reflected by partially overlapping clusters among light, medium, and dark roasted samples (Fig. 7c-d). This pattern indicates that roasting induced only moderate changes in the multivariate bioactive profile of SCGs. The observed overlap further suggests that the brewing process largely reduces compositional differences originating from roasting, resulting in more comparable SCG profiles across roasting levels.

Overall, PCA confirmed that brewing cycles were the primary factor governing the multivariate chemical profile of SCGs, whereas the roasting level acted as a secondary contributor. Similar PCA trends were observed for both varieties. Importantly, the association of SCG samples with bioactive compound loadings indicates that SCGs retain measurable levels of phenolics, flavonoids, caffeine, and CGA even after multiple brewing cycles, supporting their potential reuse in functional food applications and other value-added uses.

4. Conclusion

This study demonstrates that the chemical composition and functional potential of spent coffee grounds (SCGs) are primarily governed by processing conditions, particularly roasting degree and sequential brewing cycles, rather than coffee variety alone. Light roasting preserved higher levels of phenolics, flavonoids, chlorogenic acid, and antioxidant activity, whereas darker roasting promoted thermal degradation and reduced the extractability of these bioactive compounds. Repeated brewing progressively depleted water-soluble constituents, with the first brewing cycle accounting for the largest loss, while subsequent cycles resulted in comparatively smaller compositional changes. Although intrinsic differences between Arabica and Robusta were observed, varietal effects were less pronounced than those of roasting and brewing. Caffeine remained relatively stable across processing conditions, whereas chlorogenic acid was highly sensitive to thermal treatment and extraction history. Multivariate analysis further confirmed brewing cycles as the dominant factor shaping SCG chemical profiles, with secondary contributions from roasting intensity.

From a sustainability perspective, these findings highlight that SCGs should be regarded as a process-dependent resource rather than a uniform waste stream. This process-oriented understanding provides a practical framework for tailoring SCG valorization toward functional food ingredients and related industrial applications. Future research should focus on optimizing food-grade extraction strategies, validating SCG-derived extracts in real food or nutraceutical matrices, and assessing stability, safety, and scalability to support industrial implementation.

CRediT authorship contribution statement

Nur Maiyah: Writing – original draft, Methodology, Investigation, Data curation. Soraya Kerdpiboon: Writing – review & editing, Validation, Supervision, Resources, Project administration, Funding acquisition, Data curation, Conceptualization. William L. Kerr: Writing – review & editing. Wanwimol Klaypradit: Visualization, Validation, Data curation. Chayada Smithisukul: Methodology, Investigation, Data curation. Suriyan Supapvanich: Writing – review & editing, Supervision, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary material

Data availability

The data that has been used is confidential.