Journal Search Engine

Volume/Issue :

Year(s) : to

Search :

Title :

Author :

Keyword :

Abstract :

Figure :

Table :

Reference :

Search Advanced Search

Adode Reader(link)

View PDF Download PDF Export Citation Metrics Korean Bibliography PMC Previewer

ISSN : 1225-7060(Print)
ISSN : 2288-7148(Online)

Journal of The Korean Society of Food Culture Vol.41 No.3 pp.171-179
DOI : https://doi.org/10.7318/KJFC/2026.41.3.171

Headspace Volatile Redistribution in a Plant-based Protein Drink Supplemented with Sweet-brown Note associated Flavorings

Min Kyung Park^1,2*

, Min Jung Kim^2,3

, Jin Young Lee⁴

, Han Sub Kwak^1,2

, Min-Cheol Kang^1,2

, Dong Hyeon Park¹

¹Food Processing Research Group, Korea Food Research Institute
²Major in Food Biotechnology, University of Science and Technology
³Department of Food Biotechnology, University of Science & Technology
⁴Aromaline Co., Ltd.

^* Min Kyung Park, Food Processing Research Group, Korea Food Research Institute, Wanju, 55365, Republic of Korea Tel: +82-63-219-9582 Fax: +82-63-219-9876 E-mail: mk.park@kfri.re.kr

Received April 24, 2026 Review June 4, 2026 Accepted June 9, 2026

Abstract

This study investigated the redistribution of headspace volatile compounds in a plant-based protein drink supplemented with flavor compounds with sweet-brown note. The volatile profiles of the control (only protein drink), flavoring (F), and the protein drink supplemented with flavoring (PF) were evaluated for deviations from additive effect. Principal component analysis showed that, while some PF samples reflected flavoring characteristics, others remained close to the control, suggesting that flavor expression could not be fully explained by simple flavoring addition. A comparison of the expected and observed volatile levels confirmed non-linear behavior, with overall suppression as the dominant effect. An off-flavor-anchored redistribution map further revealed the co-suppression of most volatile compounds alongside off-flavor compounds, while a subset of alcohols and esters showed opposite trends, indicating displacement-driven release. Emergent volatile compounds were also observed, supporting the matrix-induced redistribution. These findings provide mechanistic insights into the headspace distribution for plant-based protein beverages.

Key Words : Sweet-brown note , Plant-based protein drink , Volatile compound , Flavor redistribution

Sweet-brown 향료 첨가에 따른 식물성 단백질 음료의 헤드스페이스 휘발성 향기성분의 변화 연구

박민경^1,2*, 김민정^2,3, 이진영⁴, 곽한섭^1,2, 강민철^1,2, 박동현¹

¹한국식품연구원 가공공정연구단
²과학기술연합대학원대학교 식품생명공학
³한국식품연구원 노화연구단
⁴아로마라인 주식회사

초록

키워드 :

This article has been cited by 0 article in crossref

Cited-By

Funding:

I. Introduction

Flavoring systems are widely incorporated into protein-based beverages to improve sensory quality, particularly by masking plant protein-derived off-flavors (Mittermeier-Kleßinger et al. 2021;Dai et al. 2026). Plant proteins interact with volatile compounds through hydrophobic interactions, hydrogen bonding, and physical entrapment, thereby influencing flavor release and perception (Sharma et al. 2025;Dai et al. 2026). However, most previous studies have focused on volatile retention or suppression in protein matrices, largely assuming that flavor behavior follows additive effects or equilibrium partitioning (Viry et al. 2018;Snel et al. 2023).

Protein-flavor interactions have been shown to reduce the headspace concentration of hydrophobic compounds (Jouenne & Crouzet 2001). Both protein structure and composition play a key role in volatile retention. However, how the combination of a protein matrix and added flavoring system collectively shapes the headspace volatile profile remains poorly understood. In practical applications, the perception of flavor may be governed primarily by the volatile composition of the headspace rather than the bulk concentration of volatile compounds. In complex systems, the interaction between proteins and added flavoring systems may result in non-linear effects, including selective suppression, enhancement, or the emergence of previously undetected volatile compounds (Jouenne & Crouzet 2001;Heng et al. 2004). Such interactions are particularly important to consider in the formulation of protein-based beverages, where flavor performance often deviates from expectations based on the flavoring system alone.

In this study, “sweet-brown note associated flavorings” refer to flavoring systems designed to provide sweet, caramel-like, roasted, cocoa-like, and Maillard reaction-associated sensory notes. The sweet note was represented by 4-hydroxy-2,5-dimethyl-3(2H)-furanone (furaneol)-containing flavoring, characterized by caramel-like and sweet aroma attributes, whereas the brown note was represented by cocoa-type flavorings, characterized by roasted, cocoa-like, and pyrazine-associated aroma attributes (Haag et al. 2021;Putri et al. 2024). While their sensory properties are well established, their behavior in protein matrix systems remains poorly understood, particularly regarding headspace volatile distribution and the potential emergence of novel volatile compounds.

Therefore, this study aimed to compare the headspace volatile profiles of a flavoring system with and without the addition of a plant-based protein beverage matrix. This approach allowed the evaluation of whether flavoring addition results in simple additive behavior or non-linearly alters volatile distribution in the headspace. Furthermore, the findings provide practical insights into how protein matrices influence flavor expression in formulated beverages.

II. Materials and methods

1. Materials and sample preparation

Flavoring materials with cocoa-like brown note (F_101 and F_118) and sweet note (4-hydroxy-2,5-dimethyl-3(2H)-furanone, F_Fu) were purchased from Aromaline Corp. (Seongnam-si, Gyeonggi-do, Korea). Also all other chemicals were obtained from Sigma-Aldrich (St. Louis, MO, USA).

2. Sample preparation

The model protein beverage was composed of isolated soybean protein (3.5%, w/v), minor amounts of stabilizers and buffering agents (0.02%, w/v) and water. Three sample groups were prepared: Control, containing only the plant-based protein beverage; F, containing each flavoring material in water without protein; and PF, containing the plant-based protein beverage supplemented with each flavoring material. For PF samples, flavoring materials were added to the protein beverage at 0.5% (v/v) and mixed at 400 rpm for 3 min at room temperature. The F samples were prepared using the same amount of flavoring material and final sample volume as PF, but without the protein beverage matrix. Before HS-SPME analysis, all samples were equilibrated for 30 min at room temperature to minimize differences caused by sample handling.

3. Volatile analysis using GC-MS

Volatile compounds (VOCs) were extracted using headspace solid-phase microextraction (HS-SPME) with a divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS fiber, 75 µm, Supelco). Samples (3 mL) were placed in 20 mL vials with an internal standard (100 µg/mL in methanol, 2-methyl-3-heptanone) and equilibrated at 40℃ for 30 min, followed by extraction for 30 min under agitation at 250 rpm. Analytes were thermally desorbed in the gas chromatography (GC) injector for 5 min in split-less mode.

VOCs were analyzed using a Gas chromatography-mass spectrometry (GC-MS) system (TRACE 1310 and TSQ 9000, Thermo Fisher Scientific) equipped with DB-WAX UI column (60 m × 0.25 mm × 0.25 µm; Agilent J&W Scientific, CA, USA) with helium as the carrier gas (1 mL/min). The oven temperature was programmed from 40℃ to 230℃ at 4℃/min. The inlet and transfer line were maintained at 250℃. Mass spectra were acquired in electron ionization (EI) mode over m/z 35-350. VOCs were identified using mass spectral libraries (Wiley 7.0 and NIST 14) and retention indices (C7-C30), and were semi-quantified as relative peak area ratios to the internal standard.

4. Statistical analysis

All experiments were conducted in triplicate, and results were expressed as mean ± standard deviation. Principal component analysis was performed to evaluate global differences in volatile profiles. The redistribution index (RI) was calculated as a log2-transformed ratio of observed to expected values. Directional similarity between VOCs and off-flavor compounds was quantified using cosine similarity. Data processing and visualization were conducted using Python (pandas, NumPy, and Matplotlib).

III. Results and discussion

1. Global volatile changes in headspace volatiles induced by flavoring addition

A total of 64, 59, and 51 volatile compounds were detected in the plant-based protein drink (control), flavoring only samples (F), and flavored protein drinks (PF), respectively (Table 1). In particular, the furaneol-containing flavoring (F_Fu) was characterized by distinct furanone-derived volatiles contributing to caramel-like and sweet notes, whereas the cocoa flavor systems (F_118 and F_101) were dominated by a mixture of aldehydes, ketones, and pyrazine-related compounds associated with roasted and cocoa-like attributes (Frauendorfer & Schieberle 2006). This compositional difference indicates that the F group provides a complex volatile profile with distinct chemical signatures depending on flavor type. It may contribute differently to headspace redistribution when incorporated into the protein-based matrix.

To evaluate the global changes in the volatile profile of the protein drink upon flavoring addition, principal component analysis (PCA) was performed on the VOC dataset of control, F, and PF. The PCA score plot was explained by 66.1% of the total variance (PC1: 38.3%, PC2: 27.8%), discriminating the three sample groups into distinct clusters (Figure 1). Along PC1, the control and PF groups (except PF_101) clustered on the positive dimension, suggesting that the original volatile profile of the protein drink remained dominant despite the addition of flavoring. In contrast, the position of PF_101 on the negative dimension indicated a suppressing or masking of the control volatiles. PC2 distinguished the unique flavor characteristics of F_118 and F_FU; however, only PF_101 and F_101 were positioned within the same dimension, indicating relatively close volatile relationship. This suggests that the volatile characteristics of certain flavorings may be attenuated by the matrix interactions; however, the F_101 system successfully preserved its distinct volatile characteristics within the headspace. The addition of flavoring may alter the flavor characteristics of the original system; however, the extent of such modification may be limited depending on the conditions. In a previous study, the addition of flavorings may alter beverage flavor characteristics; however, pH-responsive alginate-chitosan hydrogel matrix structure limits modification by retaining 85% limonene at pH 3.5 versus 40% release at pH 6.5 (Liu et al. 2025). These findings demonstrated that flavor expression in the protein drink is not a simple additive process, but instead reflects competitive headspace redistribution in which matrix effects and molecular interactions selectively govern the final headspace volatile profile.

To further elucidate complex volatile changes in the headspace, the relationship between expected and observed volatile levels was compared. This analysis was conducted to evaluate whether the volatile profile of PF deviated from the expected additive profile of the protein beverage and flavoring-only samples. Because the F samples did not contain the protein beverage matrix, the comparison reflects matrix-associated changes after introducing flavorings into the protein beverage system. The deviations between observed PF values and expected values were interpreted as evidence of non-additive headspace behavior under the present analytical conditions, rather than as direct quantitative measures of individual protein–volatile interactions.

The comparison between expected (control + F) and observed (PF) volatile levels revealed a clear deviation from additive behavior (Figure 2). Most volatile compounds were located below the line, indicating overall suppression of volatile release by the protein matrix. However, some volatile compounds showed enhanced release, suggesting selective matrix-driven modulation. Especially, several volatile compounds, including propan-1-ol and 3-methylbutan-1-ol, were detected in the PF groups despite being absent in both control and F samples. The detection of these VOCs in the PF groups may be attributed to matrix-induced redistribution rather than de novo formation. This pattern may be associated with matrix-induced redistribution, competitive release, altered partitioning behavior, or changes in HS-SPME extraction sensitivity, allowing compounds present below the detection limit in the individual samples to become detectable in the PF headspace (Viry et al. 2018;Snel et al. 2023). These results may indicate a complex and non-linear redistribution of volatile rather than simple additive behavior.

2. Protein matrix-responsive off-flavor control

Plant protein-derived off-flavors are primarily associated with volatile compounds formed via lipid oxidation and enzymatic reactions (Leonard et al. 2023;Dias et al. 2026). Key contributors include aldehyde (e.g. hexanal, heptanal, nonanal), alcohols (e.g. 1-octen-3-ol), furans (e.g. 2-pentylfuran), and alcohols (e.g. pentan-1-ol, hexan-1-ol) (Soendjaja & Girard 2024;Kaur et al. 2025;Nagassa et al. 2025;Dias et al. 2026). These volatile compounds are typically characterized by low odor thresholds and are responsible for undesirable sensory notes described as grassy, beany, earthy, and fatty (Jung et al. 2021;Li et al. 2023). In particular, hexanal and 2-pentylfuran are widely known as major contributors to the characteristic off-flavor of plant proteins (Park et al. 2025).

The redistribution behavior of volatile compounds in the protein drink upon flavoring addition was visualized using an off-flavor anchored scatter map (Figure 3), integrating both magnitude (mean RI) and directional similarity to soybean-derived off-flavor compounds, including hexanal, heptanal, 2-pentylfuran, pentan-1-ol, hexan-1-ol, nonanal, oct-1-en-3-ol. In Figure 3, the x-axis represents the overall matrix-driven redistribution (Mean RI), while the y-axis represents directional similarity to off-flavor compounds, enabling interpretation of VOC behavior related to off-flavor suppression. Mean RI was calculated as the average log2-transformed ratio of observed to expect volatile compounds levels across all flavoring systems, representing the overall direction of volatile redistribution {Equation (1) and (2)}. The y-axis represents the cosine similarity between the redistribution profile of each VOC and the average redistribution pattern of off-flavor compounds across flavoring systems {Equation (3)}. Suppression corresponds to VOCs that decrease together with off-flavor compounds, whereas enhancement refers to VOCs that increase while off-flavor compounds decrease, reflecting opposing redistribution trends.

RI = \log_{2} (\frac{PF}{Control + F})

(Eq. 1)

MeanRI = \frac{{RI}_{Fu} + {RI}_{118} + {RI}_{101}}{3}

(Eq. 2)

Y = \frac{\sum ({RI}_{VOC, i} \cdot {RI}_{off, i})}{\sqrt{\sum {RI}_{VOC, i}^{2}} \sqrt{\sum {RI}_{off, i}^{2}}}

(Eq. 3)

Most VOCs were located in the upper-left dimension, indicating negative RI values and strong positive similarity (Y) to the off-flavor pattern. This suggests that the majority of volatiles were co-suppressed alongside off-flavor compounds. This co-suppression may reflect matrix-associated retention of volatile compounds through hydrophobic interactions, hydrogen bonding, or physical entrapment. However, altered air–water partitioning, competitive binding among volatiles, and reduced headspace transfer may also contribute to the observed suppression. Representative co-suppressed VOCs included aldehyde (v2), methyl acetate (v6), and heptan-2-one (v29), octane (v3), α-pinene (v15), and butan-2-one (v8). A previous study reported representative co-suppressed VOCs included aldehydes, methyl acetate, 2-heptanone, octane, α-pinene, and 2-butanone, reduced by 60-85% through processing in plant protein systems (Li et al. 2023). The identified VOCs encompassed a broad range of chemical classes, including aldehydes, esters, ketones, and hydrocarbons, indicating that the observed co-suppression effect was not restricted to a specific chemical functional group but was rather governed by common physicochemical properties, particularly volatility and hydrophobicity. These findings may suggest a global suppression effect within the protein-flavor system.

In contrast, a subset of VOCs was positioned in the lower-right dimension, indicating positive RI values and negative similarity to off-flavor compounds. These compounds showed enhancement behavior while off-flavor compounds were suppressed. Specifically, several VOCs showed enhancement behavior, including butan-1-ol (v26), 3-methylbutyl acetate (v27), hexan-1-ol (v47), 3-hydroxybutan-2-one (v41), and hexyl acetate (v39), as evidenced by positive mean RI values and negative similarity to the off-flavor pattern. These compounds increased in headspace concentration despite the suppression of off-flavor components, reflecting opposing redistribution trends. These compounds included alcohols and esters with relatively higher volatility and weaker matrix affinity, suggesting preferential release upon the reduction of competing volatile interactions. Furthermore, several emergent VOCs were also observed in this dimension, suggesting matrix-induced release of previously undetected VOCs. Matrix characteristics in plant proteins can result in the detection of previously undetected volatile compounds, as mechanical disruption or competitive release of bound VOCs (Szeitz et al. 2024).

IV. Summary and conclusion

This study demonstrated that volatile behavior in a plant-based protein-flavor system was related to non-linear headspace redistribution rather than additive effects. The plant-based protein matrix induced a dominant co-suppression of volatile compounds, including key off-flavor compounds, through non-specific retention mechanisms, such as hydrophobic interactions and molecular binding. At the same time, a subset of VOCs showed opposite redistribution behavior, increasing in headspace despite off-flavor suppression, indicating displacement-driven release. Furthermore, VOCs detected only in the combined system (PF) suggest that the protein-flavor matrix may increase the headspace availability of compounds that were previously undetectable, possibly through redistribution, competitive release, or altered partitioning behavior.

These findings suggest that volatile distribution is strongly influenced by the protein beverage matrix, providing a useful basis for understanding off-flavor modulation in plant-based protein beverages. Also, it highlights that volatile distribution in protein-based beverage system is not governed by simple partitioning or additive effects, but rather by complex interaction-driven mechanisms. From a practical perspective, this suggests that flavor profile in plant-based protein beverages cannot be accurately predicted based solely on the properties of the flavoring system. Instead, formulation strategies should consider matrix-dependent interactions to achieve desired sensory properties. Although RI and directional similarity provided quantitative support for classifying volatile redistribution patterns, these analyses indicate association rather than direct molecular causation. Further validation using targeted protein–volatile binding assays or time-resolved headspace analysis would be required to confirm the proposed mechanisms. It should also be noted that the VOC data were based on semi-quantification using relative peak area ratios to an internal standard. Because HS-SPME fiber affinity can differ among compounds, the present data should be interpreted primarily for comparing redistribution trends within the same compound and analytical workflow, rather than for direct absolute comparison among different compounds.

Therefore, this study provided a mechanistic framework for understanding headspace volatile behavior in protein-based systems and offered useful insights for the rational design of flavor systems, contributing to the development of improved plant-based beverages with enhanced sensory quality.

Acknowledgement

This study was supported by the Main Research Program [E0232201] of the Korea Food Research Institute funded by the Ministry of Science and ICT (Republic of Korea) and the High Value-Added Food Technology Development Program [RS-2026-25524143] of the Korea Institute of Planning and Evaluation for Technology in Food, Agriculture, and Forestry (IPET), funded by Ministry of Agriculture, Food and Rural Affairs (Republic of Korea).

Author biography

Min Kyung Park (Korea Food Research Institute, https://orcid.org/0000-0002-3619-9491)

Min Jung Kim (Korea Food Research Institute, https://orcid.org/0000-0003-0205-016X)

Jin Young Lee (Aromaline Co., Ltd., https://orcid.org/0000-0004-4578-4184)

Han Sub Kwak (Korea Food Research Institute, https://orcid.org/0009-0003-4270-6821)

Min-Cheol Kang (Korea Food Research Institute, https://orcid.org/0000-0002-9658-9045)

Dong Hyeon Park (Korea Food Research Institute, https://orcid.org/0000-0002-6831-8854)

Conflict of Interest

No potential conflict of interest relevant to this article was reported.

Figure

<Figure 1>.

The Score plot of principal component analysis based on volatile compounds in headspace. All abbreviations are follows: Control, protein drink only; PF, flavored protein drink; F, flavoring.

<Figure 2>.

Parity plot (observed vs. expected) of volatile compounds showing deviation from flavoring addition in the protein-flavor system. All abbreviations are follows: PF, flavored protein drink; F, flavoring.

<Figure 3>.

Off-flavor anchored redistribution of volatile compounds based on mean redistribution index (mean RI) and directional similarity. (a) Redistribution map showing the relationship between mean RI and directional similarity to off-flavor compounds. (b) Representative volatile compounds classified according to co-suppression and opposite redistribution behavior.

Table

<Table 1>.

Volatile compounds identified by GC-MS

All abbreviations are follows: Control, protein drink only; PF, flavored protein drink; F, flavoring.

^aNumbered in the order of retention indices (RI).

^bRetention indices (RI) were determined using n-alkanes (C7-C22).

^cMean values of relative peak area to that of internal standard ± standard deviation.

^dNot detected

No.a	Volatile compound	RIb	Control	F_101	F_118	F_Fu	PF_101	PF_118	PF_FU
			Contentsc (Mean±SD)

Acids
v55	acetic acid	1457	1.01 ± 0.09	0.26 ± 0.16	8.45 ± 3.27	23.09 ± 3.0	NDd)	ND	1.85 ± 0.25
v66	hexanoic acid	>1800	0.71 ± 0.46	0.17 ± 0.09	4.42 ± 1.18	0.13 ± 0.03	0.1 ± 0.06	0.1 ± 0.04	0.05 ± 0.0
v72	octanoic acid	>1800	0.11 ± 0.02	0.03 ± 0.01	0.62 ± 0.12	0.27 ± 0.12	0.06 ± 0.03	0.03 ± 0.02	ND
v75	benzoic acid	>1800	3.03 ± 0.34	ND	ND	ND	ND	ND	ND
Aldehydes
v2	acetaldehyde	<800	1.38 ± 0.61	0.06 ± 0.02	1.64 ± 0.29	2.15 ± 0.69	0.04 ± 0.01	1.35 ± 0.3	0.56 ± 0.08
v4	2-methylpropanal	812	0.25 ± 0.04	0.02 ± 0.0	ND	ND	ND	0.27 ± 0.05	0.05 ± 0.01
v7	butanal	858	0.43 ± 0.03	0.01 ± 0.0	0.48 ± 0.07	0.05 ± 0.01	0.04 ± 0.01	0.1 ± 0.03	0.09 ± 0.05
v9	2-methylbutanal	908	0.37 ± 0.17	ND	ND	ND	ND	0.4 ± 0.02	0.07 ± 0.02
v10	3-methylbutanal	912	0.49 ± 0.07	0.49 ± 0.06	0.63 ± 0.11	0.07 ± 0.03	0.04 ± 0.01	0.64 ± 0.07	0.31 ± 0.09
v20	Hexanal	1088	37.33 ± 4.36	0.66 ± 0.13	0.47 ± 0.07	0.06 ± 0.03	0.07 ± 0.02	2.35 ± 1.32	0.62 ± 0.13
v30	heptanal	1192	1.05 ± 0.07	ND	ND	ND	ND	0.04 ± 0.01	ND
v42	(E)-2-ethylhex-2-enal	1336	0.07 ± 0.02	ND	ND	ND	ND	0.05 ± 0.01	0.03 ± 0.01
v50	nonanal	1402	0.88 ± 0.24	ND	ND	0.04 ± 0.01	ND	0.04 ± 0.01	0.02 ± 0.01
Alcohols
v11	ethanol	928	0.44 ± 0.11	8.37 ± 2.49	6.55 ± 1.71	2.25 ± 0.57	0.54 ± 0.03	1.64 ± 0.98	1.76 ± 0.56
v17	propan-1-ol	1032	ND	ND	ND	ND	7.79 ± 1.63	4.35 ± 0.62	4.57 ± 2.26
v26	butan-1-ol	1146	0.25 ± 0.1	ND	ND	ND	0.4 ± 0.06	0.39 ± 0.11	0.37 ± 0.09
v33	3-methylbutan-1-ol	1220	ND	ND	ND	ND	0.2 ± 0.06	0.21 ± 0.06	0.2 ± 0.09
v37	pentan-1-ol	1262	3.1 ± 0.81	0.05 ± 0.02	ND	0.05 ± 0.03	0.27 ± 0.06	5.49 ± 1.52	3.49 ± 1.15
v47	hexan-1-ol	1388	0.58 ± 0.4	0.03 ± 0.01	ND	ND	3.16 ± 0.35	18.51 ± 4.27	12.91 ± 3.54
v54	oct-1-en-3-ol	1449	1.89 ± 0.43	0.02 ± 0.01	ND	ND	0.42 ± 0.15	1.83 ± 0.1	1.24 ± 0.37
v56	heptan-1-ol	1471	0.33 ± 0.1	ND	ND	ND	0.11 ± 0.01	0.57 ± 0.22	0.39 ± 0.14
v59	octan-1-ol	1523	0.22 ± 0.01	ND	ND	ND	0.09 ± 0.04	0.27 ± 0.09	0.18 ± 0.05
Aliphatic hydrocarbons
v1	heptane	<800	0.69 ± 0.11	ND	ND	ND	0.01 ± 0.0	0.68 ± 0.34	0.26 ± 0.1
v3	octane	800	1.55 ± 0.41	0.04 ± 0.01	3.99 ± 0.66	0.46 ± 0.02	0.05 ± 0.02	0.98 ± 0.2	0.47 ± 0.03
v14	decane	1000	0.17 ± 0.05	0.06 ± 0.02	6.07 ± 0.51	1.38 ± 0.11	0.08 ± 0.05	0.08 ± 0.03	0.08 ± 0.02
v21	undecane	1100	0.12 ± 0.04	0.23 ± 0.08	19.53 ± 3.11	2.26 ± 0.34	ND	ND	ND
Benzenes and derivatives
v16	toluene	1026	0.84 ± 0.32	0.03 ± 0.0	2.1 ± 0.77	0.08 ± 0.01	0.03 ± 0.01	0.55 ± 0.1	0.34 ± 0.14
v23	1,2-xylene	1126	0.4 ± 0.14	ND	ND	ND	0.04 ± 0.01	0.32 ± 0.19	0.2 ± 0.04
v25	1,4-xylene	1138	0.38 ± 0.12	ND	ND	ND	0.03 ± 0.01	0.35 ± 0.09	0.24 ± 0.07
v28	1,3-xylene	1162	0.39 ± 0.05	0.01 ± 0.01	2.78 ± 0.48	ND	0.03 ± 0.0	0.33 ± 0.02	0.21 ± 0.08
v61	2-phenylacetaldehyde	1624	0.06 ± 0.01	ND	ND	ND	ND	ND	ND
v64	ethyl 2-phenylacetate	1483	0.04 ± 0.01	ND	ND	ND	ND	ND	ND
v67	phenylmethanol	>1800	0.1 ± 0.02	0.01 ± 0.0	0.24 ± 0.1	0.03 ± 0.01	0.02 ± 0.0	0.05 ± 0.02	0.04 ± 0.03
v68	2-phenylethanol	>1800	0.07 ± 0.02	ND	ND	ND	ND	ND	ND
Esters
v6	methyl acetate	832	0.04 ± 0.01	ND	ND	3.26 ± 0.21	ND	0.05 ± 0.02	0.05 ± 0.0
v13	propyl acetate	980	3.19 ± 0.65	0.24 ± 0.05	1.36 ± 0.32	9.18 ± 3.36	16.45 ± 0.62	3.0 ± 1.0	3.45 ± 1.1
v27	3-methylbutyl acetate	1158	0.17 ± 0.01	0.02 ± 0.01	ND	0.03 ± 0.01	1.31 ± 0.49	0.52 ± 0.22	0.71 ± 0.13
v35	ethyl hexanoate	1231	0.3 ± 0.03	0.08 ± 0.04	1.24 ± 0.2	0.58 ± 0.27	0.04 ± 0.01	0.25 ± 0.1	0.19 ± 0.04
v39	Hexyl acetate	1288	0.08 ± 0.01	0.02 ± 0.01	1.32 ± 0.49	1.64 ± 0.8	3.71 ± 1.1	0.68 ± 0.38	1.39 ± 0.44
v52	ethyl octanoate	1424	ND	0.09 ± 0.03	ND	0.03 ± 0.01	0.13 ± 0.05	ND	ND
Furans and derivatives
v12	2-ethylfuran	946	3.32 ± 0.35	ND	ND	ND	0.07 ± 0.02	3.05 ± 0.52	1.83 ± 0.35
v24	2-butylfuran	1132	0.54 ± 0.15	ND	ND	ND	0.03 ± 0.01	0.27 ± 0.12	0.13 ± 0.02
v34	2-pentylfuran	1228	13.02 ± 3.72	ND	ND	ND	1.28 ± 0.42	7.13 ± 1.99	2.65 ± 0.52
v57	furan-2-carbaldehyde	1488	0.23 ± 0.13	ND	0.69 ± 0.14	ND	0.01 ± 0.01	0.06 ± 0.02	ND
v63	S-(furan-2-ylmethyl) ethanethioate (Furfurylthiol acetate)	1769	ND	0.09 ± 0.03	ND	ND	0.12 ± 0.01	ND	ND
v71	4-hydroxy-2,5-dimethylfuran-3-one (furaneol)	>1800	0.06 ± 0.02	0.05 ± 0.02	0.52 ± 0.16	17.08 ± 5.16	ND	ND	16.32 ± 4.95
Ketones
v5	propan-2-one	826	5.09 ± 0.91	0.12 ± 0.02	5.31 ± 2.19	2.66 ± 0.14	0.76 ± 0.17	6.26 ± 1.68	3.45 ± 1.6
v8	butan-2-one	902	4.2 ± 0.39	0.02 ± 0.01	1.74 ± 0.42	2.5 ± 0.83	0.1 ± 0.01	6.07 ± 1.84	3.07 ± 0.58
v18	pentane-2,3-dione	1042	0.21 ± 0.03	0.01 ± 0.0	ND	1.54 ± 0.39	0.06 ± 0.02	0.26 ± 0.12	0.44 ± 0.07
v29	heptan-2-one	1182	4.35 ± 2.42	0.01 ± 0.01	0.33 ± 0.27	0.03 ± 0.01	0.29 ± 0.07	2.41 ± 1.32	1.46 ± 0.41
v40	octan-2-one	1292	0.23 ± 0.08	0.01 ± 0.0	ND	ND	0.16 ± 0.02	0.48 ± 0.03	0.39 ± 0.07
v41	3-hydroxybutan-2-one	1302	0.45 ± 0.09	0.82 ± 0.12	ND	0.25 ± 0.07	0.32 ± 0.06	9.93 ± 1.57	17.4 ± 1.71
v43	1-hydroxypropan-2-one	1342	0.23 ± 0.06	0.41 ± 0.07	26.21 ± 9.56	26.65 ± 12.36	3.44 ± 1.15	10.49 ± 4.8	12.53 ± 2.35
Lactones
v62	5-ethyloxolan-2-one (γ-Hexalactone)	1694	ND	0.01 ± 0.01	0.87 ± 0.11	0.06 ± 0.02	ND	ND	ND
v65	2-hydroxy-3-methylcyclopent-2-en-1-one (Maple lactone, Corylone)	>1800	ND	0.27 ± 0.06	ND	ND	0.02 ± 0.01	ND	ND
v73	5-hexyloxolan-2-one (γ-decalactone)	>1800	ND	0.49 ± 0.15	ND	0.05 ± 0.03	0.74 ± 0.28	ND	ND
v74	6-pentyloxan-2-one (δ-decalactone)	>1800	ND	0.11 ± 0.04	ND	ND	0.21 ± 0.02	ND	ND
Pyrazines
v38	2-methylpyrazine	1284	0.81 ± 0.08	4.06 ± 1.04	ND	ND	1.26 ± 0.57	0.92 ± 0.04	0.64 ± 0.28
v44	2,5-dimethylpyrazine	1357	0.18 ± 0.05	4.46 ± 2.88	ND	ND	2.42 ± 0.57	0.2 ± 0.02	0.13 ± 0.03
v45	2,6-dimethylpyrazine	1366	0.13 ± 0.03	0.67 ± 0.08	ND	ND	0.35 ± 0.1	0.16 ± 0.04	0.12 ± 0.01
v46	2,3-dimethylpyrazine	1383	0.1 ± 0.03	0.37 ± 0.05	ND	ND	0.19 ± 0.07	0.11 ± 0.03	0.08 ± 0.01
v49	2-ethyl-6-methylpyrazine	1398	0.11 ± 0.02	0.78 ± 0.3	ND	ND	0.79 ± 0.26	0.11 ± 0.03	0.08 ± 0.01
v51	2,3,5-trimethylpyrazine	1407	0.17 ± 0.04	0.79 ± 0.18	ND	ND	0.61 ± 0.27	0.2 ± 0.04	0.14 ± 0.03
v53	3-ethyl-2,5-dimethylpyrazine	1430	0.04 ± 0.01	0.59 ± 0.04	ND	ND	0.65 ± 0.08	0.05 ± 0.02	0.03 ± 0.0
v58	2,3,5,6-tetramethylpyrazine	1494	0.04 ± 0.01	ND	ND	ND	0.01 ± 0.0	0.05 ± 0.01	0.03 ± 0.01
v60	1-pyrazin-2-ylethanone	1607	ND	70.87 ± 17.29	ND	ND	41.83 ± 19.09	ND	ND
Sulfur-containing compounds
v19	(methyldisulfanyl)methane	1068	0.56 ± 0.05	ND	ND	ND	0.01 ± 0.0	1.99 ± 0.45	1.42 ± 0.24
v32	S-methyl butanethioate	1215	0.05 ± 0.02	2.99 ± 0.94	0.93 ± 0.2	2.09 ± 0.54	8.53 ± 3.74	0.04 ± 0.01	0.03 ± 0.01
v48	(methyltrisulfanyl)methane	1392	0.08 ± 0.02	ND	ND	ND	ND	0.16 ± 0.03	0.03 ± 0.0
Terpenes
v15	2,6,6-trimethylbicyclo[3.1.1]hept-2-ene (a-pinene)	1015	2.67 ± 1.04	ND	ND	ND	0.1 ± 0.04	2.72 ± 0.63	0.99 ± 0.11
v22	6,6-dimethyl-2-methylidenebicyclo[3.1.1]heptane (beta-pinene)	1102	0.36 ± 0.08	ND	ND	ND	0.02 ± 0.01	0.32 ± 0.08	0.12 ± 0.01
v31	1-methyl-4-prop-1-en-2-ylcyclohexene	1196	0.2 ± 0.07	ND	ND	ND	0.26 ± 0.02	0.12 ± 0.05	0.08 ± 0.02
v36	1-methyl-4-propan-2-ylcyclohexa-1,4-diene (gamma-terpinene)	1244	ND	0.01 ± 0.01	1.5 ± 0.21	ND	0.01 ± 0.0	ND	ND
Miscellaneous compounds
v69	3-hydroxy-2-methylpyran-4-one (maltol)	>1800	0.08 ± 0.03	0.33 ± 0.04	ND	ND	0.01 ± 0.0	ND	ND
v70	2-ethyl-3-hydroxypyran-4-one (ethyl maltol)	>1800	ND	0.64 ± 0.14	ND	ND	0.11 ± 0.06	ND	ND

Reference

Dai, X.,Zhang, J.,Chen, Z.,Awais, M.,Fan, B.,Fauconnier, M. L.,Liu, L.,Wang, F. 2026. Protein-flavor interactions in plant protein food matrix: molecular binding mechanisms, influencing factors, and modulation strategies . Compr Rev Food Sci Food Saf, 25(1), e70318.
Dias, F. F. G.,Oliveira, W. d. S.,Bogusz Junior, S.,Taha, A. Y. 2026. Integrated assessment of lipid oxidation and off-flavor formation in plant protein powders . J Am Oil Chem Soc. 103(5), 371-386.
Frauendorfer, F.,Schieberle, P. 2006. Identification of the key aroma compounds in cocoa powder based on molecular sensory correlations . J Agric Food Chem, 54(15), 5521-5529.
Haag, F.,Hoffmann, S.,Krautwurst, D. 2021. Key food furanones furaneol and sotolone specifically activate distinct odorant receptors . J Agric Food Chem, 69(37), 10999-11005.
Heng, L.,Van Koningsveld, G.,Gruppen, H.,Van Boekel, M.,Vincken, J.-P.,Roozen, J.,Voragen, A. 2004. Protein-flavour interactions in relation to development of novel protein foods . Trends Food Sci Technol, 15(3-4), 217-224.
Jouenne, E.,Crouzet, J. 2001. Aroma compounds—proteins interaction using headspace techniques . Headspace Analysis of Foods and Flavors: Theory and Practice, 33-41.
Jung, H.,Kim, I.,Jung, S.,Lee, J. 2021. Oxidative stability of chia seed oil and flax seed oil and impact of rosemary (Rosmarinus officinalis L.) and garlic (Allium cepa L.) extracts on the prevention of lipid oxidation . Appl Biol Chem, 64(1), 6.
Kaur, M.,Gray, C.,Barringer, S. 2025. Controlling off-odors in plant proteins using sequential fermentation . Foods, 15(1), 39.
Leonard, W.,Zhang, P.,Ying, D.,Fang, Z. 2023. Surmounting the off-flavor challenge in plant-based foods . Crit Rev Food Sci Nutr, 63(30), 10585-10606.
Li, X.,Zhang, W.,Zeng, X.,Xi, Y.,Li, Y.,Hui, B.,Li, J. 2023. Characterization of the major odor-active off-flavor compounds in normal and lipoxygenase-lacking soy protein isolates by sensory-directed flavor analysis . J Agric Food Chem, 71(21), 8129-8139.
Liu, X.,Yu, L.,Fang, Y.,Zhang, W.,Li, G.,Zeng, X.,Zhang, Y. 2025. Construction and controlled flavor release of high internal phase emulsion stabilized by pH-driven-assembled soy peptide nanoparticles . Food Chem, 471, 142806.
Mittermeier-Kleßinger, V. K.,Hofmann, T.,Dawid, C. 2021. Mitigating off-flavors of plant-based proteins . J Agric Food Chem, 69(32), 9202-9207.
Nagassa, M.,He, S.,Liu, S.,Luo, S.,Li, X.,Wu, Z.,Song, J.,Sun, H. 2025. The development of volatile off-flavor compounds in soy protein isolates and plant meat during storage . Food Chem, 481, 144025.
Park, G.-W.,Park, K.-H.,Kim, S.-G.,Lee, S.-Y. 2025. Profiles of aroma volatile components in textured vegetable proteins using headspace solid phase microextraction-gas chromatography-mass spectrometry . Curr Res Food Sci, 10, 100999.
Putri, D. N.,De Steur, H.,Juvinal, J. G.,Gellynck, X.,Schouteten, J. J. 2024. Sensory attributes of fine flavor cocoa beans and chocolate: A systematic literature review . J Food Sci, 89(4), 1917-1943.
Sharma, B.,Keast, R.,Liem, D. G.,Nolvachai, Y.,Gamlath, S.,Oliver, P.,Costanzo, A. 2025. Plant-protein isolates and flavour perception: Understanding mechanisms and strategies to balance flavour retention and release . Food Chem, 493, 145815.
Snel, S. J.,Pascu, M.,Bodnár, I.,Avison, S.,van der Goot, A. J.,Beyrer, M. 2023. Flavor-protein interactions for four plant proteins with ketones and esters . Heliyon, 9(6).
Soendjaja, V.,Girard, A. L. 2024. Effects of plant polyphenols on lipid oxidation in pea and soy protein solutions . Food Chem, 433, 137340.
Szeitz, A.,Sutton, A. G.,Hallam, S. J. 2024. A matrix-centered view of mass spectrometry platform innovation for volatilome research . Front Mol Biosci, 11, 1421330.
Viry, O.,Boom, R.,Avison, S.,Pascu, M.,Bodnár, I. 2018. A predictive model for flavor partitioning and protein-flavor interactions in fat-free dairy protein solutions . Food Res Int, 109, 52-58.