Impact of compression methods on flavor profile of white tea: Integrated analysis of appearance, aroma, and taste

Abstract

Shaping significantly influences the flavor profile of white tea. This study employed integrated metabolomics-chemometrics to analyze non-volatile/volatile compounds in three processing protocols: compressing withered leaves into tea cakes (WT), lightly compressing into tea cakes after steaming (LS), and heavily compressing into tea cakes after steaming (HS), with traditional white tea (loose tea) (TT) as the control group. The findings indicated that WT exhibited significantly elevated contents of catechins and total amino acids compared to LS and HS as well as greater soluble sugar content than LS. Nine key volatile compounds were identified including: nonanoic acid, 1-octen-3-ol, (E, E)-3,5-octadien-2-one, and 4-oxoisophoron. The analysis of non-volatile and volatile key compounds aligned with the sensory evaluations of sweetness, and floral-fruit aromas. The WT technology produced appealing fruity and floral aromas while preserving freshness and sweetness, offering a new approach for white tea cake processing.

Highlights

• •Different compressing methods had significant impacts on white tea flavor profiles.
• •Withered white tea cakes enhanced floral, fruity aromas, reducing grassy ones.
• •WT exhibited enhanced freshness and sweetness of the tea cakes.
• •The total amino acids, fresh amino acids, and polyphenols were all increased by WT.
• •This technique offered an alternative for portable production with premium quality.

1. Introduction

White tea is renowned for its natural appearance of buds and leaves, characterized by pekoe and a floral, refreshing flavor. The main production occurs in Fuding, Zhenghe, Songxi, and Jianyang in Fujian Province, China, regions extensively studied by Zhou et al. (2023) for their unique terroir and manufacturing practices that define the geographical identity and quality attributes of white tea. Recent years have seen an increase in scholarly interest in “white tea fever”, with researchers examining the effects of withering, drying, and other processes on the physicochemical quality changes of traditional white tea. Additionally, the emergence of white tea cake has significantly enhanced the variety of white tea products. Traditional white tea is primarily stored and transported in a loose form. However, its loose structure, voluminous nature, and susceptibility to breakage during transit elevate transportation, packaging, and storage costs. The compressed cake form of white tea effectively addresses various storage and transportation challenges associated with loose-leaf tea, establishing compressed white tea as a significant product category in the industry. At present, the national standard GB/T 31751–2015 “Compressed White Tea” has been formulated.

Traditional manufacturing processes typically involve high-temperature steam softening of loose tea leaves followed by high-pressure compression to achieve the desired shaping (Feng, Tan, et al., 2024). However, this conventional method significantly diminishes the characteristic fresh and brisk flavor profile intrinsic to white tea (Wang et al., 2023). Current technical frameworks for white tea compression remain underdeveloped, with existing processes predominantly adapted from dark tea compression techniques and a notable scarcity of systematic research. To address these limitations, this study focuses on developing an innovative compression method designed to maximally preserve the original flavor characteristics of loose-leaf white tea.

This study explores the influence of different tea cake compression techniques on the shape and flavor profile evolution of white tea cakes by analyzing variations in non-volatile and volatile compounds through metabolomics.

2. Materials and methods

2.1. Tea samples treatment

The experiment utilized secondary Baimudan tea, provided by Zhenghe Ruiming Tea Co., Ltd. The fresh tea leaves underwent the process of withering in an air-withering room, maintained at a temperature of 28 ± 1 °C and a relative humidity of 65 ± 5 %. Upon reaching a leaf moisture content of 15 % ± 1 % (Fang et al., 2022; Xiang et al., 2024), various cake pressing methods were employed for treatment (Fig. 1A). Three independent biological replicates were analyzed per experimental condition to ensure statistical robustness. Cake pressing equipment and method: The self-developed new-type cake press machine (6CYB- White Tea Cake Press, pressure 15 ± 1 kPa) and the traditional cake press machine (XFC-2GY- White Tea Cake Press, pressure 60 ± 1 kPa) were employed. The pressure parameters of both machines were user-adjustable, and the cake pressing method was detailed in Table 1.

Fig. 1.

(A) Flowchart of the preparation process for four tea samples. (B) Radar plot illustrating the characteristic aroma profile of four white tea samples. (C) Radar plot depicting the characteristic taste profile of four white tea samples. (D) Volume plots of four tea samples. TT: Traditional white tea. WT: Compressing withered leaves into tea cakes. LS: Lightly compressing into tea cakes after steaming. HS: Heavily compressing into tea cakes after steaming.

Table 1.

Different cake-pressing treatment methods.

Sample Name	Treatment	Remarks
TT	Traditional white tea (loose tea)	The leaves were withered at 80 °C and dried for 2 h until the moisture content was less than 7 %.
WT	Compressing withered leaves into tea cakes	The withered leaves were lightly pressed (15 ± 1 kPa, 6CYB- White Tea Cake Press) into cakes and then dried at 80 °C until their moisture content dropped below 7 %.
LS	Lightly compressing into tea cakes after steaming	The coarse tea was steamed at 100 °C for 150 s, lightly pressed (15 ± 1 kPa, 6CYB- White Tea Cake Press) into cakes, and subsequently dried at 80 °C until the moisture content dropped below 7 %.
HS	Heavily compressing into tea cakes after steaming	The coarse tea was steamed at 100 °C for 150 s, then heavily pressed (60 ± 1 kPa, XFC-2GY- White Tea Cake Press) into cakes, and dried at 80 °C until the moisture content was less than 7 %.

2.2. Sensory evaluation merged with quantitative descriptive analysis (QDA)

We utilized QDA to illustrate the odor profiles of tea samples in this research. Refer to GB/T 23776–2018, titled “Tea Sensory Evaluation Methods”. A quantity of three grams of tea was placed into a 150-mL cup, followed by the addition of boiling water for a duration of 5 min to assess the aroma and taste. The evaluation panel included five professionals involved in tea production and sensory evaluation, consisting of three males and two females. To describe the aroma of tea samples, six attributes were employed: floral and fruity, sweet aroma, grass aroma, pekoe aroma, strong, and fragrant and lasting. For the taste characterization, five attributes were used: sweet, fresh, pekoe, astringent, and mellow and thick. Panelists evaluated the intensity of each attribute using a 10-point scale ranging from 1 (non-existent) to 10 (extremely intense). Following the collection of average scores for each attribute, the data were utilized to create the chart. Throughout the study, protocols were enforced to safeguard the rights and confidentiality of all involved participants. Before the experiment, all group members were notified, and written informed consent was obtained from each participant.

2.3. Chemicals and reagents

Sigma-Aldrich (St. Louis, MO, USA) supplied a range of compounds with a purity of ≥95 %, including methanol, formic acid, acetonitrile, catechin (C), epigallocatechin gallate (EGCG), epigallocatechin (GC), epigallocatechin (EGC), epicatechin gallate (ECG), epicatechin (EC), methylation of epigallocatechin gallate (EGCG-CH₃) (all with a purity of ≥95 %), γ-aminobutyric acid (GABA), tryptophan (Trp), glutamate (Glu), lysine (Lys), proline (Pro), tyrosine (Tyr), arginine (Arg), valine (Val), alanine (Ala), leucine (Leu), isoleucine (Ile), glutamine (Gln), serine (Ser), histidine (His), asparagine (Asn), cysteine (Cys), aspartic acid (Asp), phenylalanine (Phe), glycine (Gly), β-alanine (β-Ala), theanine (Thea) (all with a purity of ≥95 %), glucose, fructose, galactose, sucrose, fucose, lactose, raffinose, melezitose, ribitol, and N, O-Bis (Trimethylsilyl) trifluoroacetamide (BSTFA) (all chromatography grade). Deionized water was produced using a Milli-Q purification system.

2.4. A method utilizing ultra-performance liquid chromatography coupled with triple quadrupole mass spectrometry (UPLC-QqQ MS) was determined for amino acids and catechins

The methodologies for the measurement of amino acids and catechins were performed based on previous studies (Chen et al., 2018, Chen, Liu, et al., 2020). For UPLC-QqQ MS analysis, a 2 mL centrifuge tube was filled with 0.3 g of the finely powdered sample, and subsequently, 1 mL of 70 % methanol was added into the tube. The mixture underwent sonication for a duration of 20 min at room temperature, subsequently being centrifuged at 4 °C with a rotational speed of 12,000 rpm for 10 min. The supernatant was extracted to be used as the stock solution. For amino acid analysis, the stock solution was diluted by a factor of 100, whereas for catechin analysis, it was diluted by a factor of 400, with triplicate measurements performed for each sample.

A quantitative analysis of amino acids and catechins was conducted using the Waters Acquity UPLC system, incorporating the advanced ultra-performance liquid chromatography-triple quadrupole mass spectrometry (UPLC-TQS MS) technology, in combination with a photodiode array (PDA) detector and the Waters-manufactured XEVO TQ-S triple quadrupole mass spectrometer, sourced from Milford, MA, USA. The sample chamber was kept at 10 °C, and dilution was performed within the range of the calibration curve. A column by Merck the ZIC-pHILIC with dimensions (100 mm × 2.1 mm, 5 μm) was utilized for amino acid content determination. The eluents comprised 0.1 % formic acid in acetonitrile and 5 mM ammonium acetate in water, with a flow rate of 0.4 mL/min. The column temperature was set at 40 °C, and an injection volume of 2 μL. The Waters Acquity UPLC BEH C18 column (100 mm × 2.1 mm, 1.7 μm), was employed for the analysis of catechins. A solvent system consisting of 0.1 % formic acid in water and 0.1 % formic acid in acetonitrile was used for elution, with a flow rate of 0.3 mL/min and a column temperature set at 40 °C, with an injection volume of 1 μL. Multireaction monitoring was refined for quantification, and calibration curves were established to ascertain concentrations. The standard calibration curves for all amino acids and catechins are provided in Supplementary Figs. S1 and S2.

2.5. UPLC-TQS-MS method for determination of soluble sugars

The measurement of soluble sugars was performed using the methods established by Wang et al. (Wang et al., 2021). To measure the concentration of soluble sugars, a 0.30 g of portion the ground sample was added to a 2 mL centrifuge tube, followed by the addition of 480 μL of a methanol-water solvent mixture (volume ratio 3:1), along with 20 μL of a vanillic acid solution containing 5 mg/mL. The mixture underwent ultrasonication for a duration of 30 min at room temperature. Subsequently, the sample was centrifuged at 12000 rpm for 10 min at a temperature of 4 °C. 150 μL of the supernatant was transferred to a 2 mL sample tube and dried under vacuum conditions. Following the complete evaporation of water, 80 μL of a 20 mg/mL methoxypyridine solution was added and held at 80 °C for 20 min. Afterward, 80 μL of BSTFA containing 1 % TMCS was added and kept at 70 °C for an hour. Following filtration, the sample was transferred to an insert tube for analysis.

The content of soluble sugars was analyzed through the use of an Agilent 7890B gas chromatograph, which was configured with a GERSTEL MPS automated sampler and an LECO Pegasus HT time-of-flight mass spectrometer (TOF-MS). The Restek Rxi®-5 Sil MS capillary column, measuring 30 m × 0.25 mm × 0.25 μm, was employed in the mass spectrometer. During the splitless injection mode, the injection port temperature was set at 280 °C, while the column flow rate of 1.5 mL/min, and the temperature of the transfer line was maintained at 275 °C. The oven temperature protocol consisted of the following steps: initiate at 80 °C for 0.2 min, increase to 160 °C at a speed of 15 °C/min for 0 min, elevate to 200 °C at a speed of 3 °C/min for 0 min, and ultimately reach 300 °C at a speed of 10 °C/min for 8 min. The mass spectrometer operated within a mass-to-charge ratio range of 35 to 600 amu, acquiring data at a rate of 10 spectra per second, a detector voltage set at 1550 V, an ion source of 250 °C, and an electron impact energy of −70 eV for analysis. Each sample underwent triplicate analysis.

2.6. SPME -GC–MS method for determination of volatile metabolite

Established methodologies were adhered to for the extraction and quantitation of tea aroma components (Chen et al., 2021; Peng et al., 2024). A 20 mL headspace vial was filled with a 2 g sample of ground tea, the vial containing 1.5 μL of decanoate ethyl ester (internal standard, 112 μg/kg). A polydimethylsiloxane/divinylbenzene (PDMS/DVB, 65 μm) fiber (Supelco, Bellefonte, PA, USA) for SPME with an incubation at 80 °C lasting 31 min, and a 60-min extraction, followed by analysis at a fiber pre-bake temperature of 250 °C for 3.5 min.

The analysis of volatile compounds was executed utilizing the method proposed by Hao et al. (Hao et al., 2023). A gas chromatograph, model 7890B, sourced from Agilent Co., was interfaced with a Pegasus HT time-of-flight mass spectrometer (GC-TOF-MS), produced by LECO Corporation based in Saint Joseph, Michigan, USA. A Restek Rxi®-5 Sil MS capillary column (30 m × 0.25 mm × 0.25 μm) film thickness was employed. Helium served as the carrier gas, flowing at a rate of 1 mL/min. A volume of 1 μL was injected in splitless mode, with the injection temperature set at 250 °C. The temperature of the transfer line was established at 275 °C. The temperature program commenced at 50 °C for 5 min, followed by an increase from 50 °C to 210 °C at a speed of 3 °C/min over 3 min, and then elevated to 230 °C at a speed of 15 °C/min. During the mass spectrometry analysis, an electron ionization (EI) energy of 70 eV, a solvent delay of 5 min, and an ion source temperature was maintained at 250 °C, and a collection speed of 10 spectra per second. The mass scan range extended from 30 to 500 m/z over a duration of 200 s.

Volatile compounds were identified by matching their peaks with the NIST mass spectrometry database. Using n-alkanes (C₈–C₃₀) as reference standards, retention indices (RI) were computed and were subsequently compared to the RI values in the NIST WebBook (http://webbook.nist.gov/ chemistry/) for confirmation. The chemical structures, names, and CAS numbers were obtained from ChemicalBook (https://www.chemicalbook.com). The odor descriptions used were obtained from the Flavor Ingredient Library (https://www.femaflavor.org/flavorlibrary/). Relative compound contents were calculated by dividing the areas of individual peaks by the total peak area, following established methodologies (Xiao et al., 2022).

2.7. ROAV calculation methods

The calculation of relative odor activity values (ROAV) for volatile components by prior studies (Fang et al., 2022; Liang et al., 2023). The component with the greatest flavor contribution in the tea sample received an ROAV of 100, while other compounds were calculated using the specified equation. Components with an ROAV of 1 or greater were identified as primary flavor contributors, whereas those with an ROAV between 0.1 and 1 were deemed to contribute to the overall flavor profile(Li et al., 2024; Su et al., 2022):

ROAV_i = $\frac{{\mathrm{C}}_{\mathrm{i}}}{{\mathrm{C}}_{\max }}\times \frac{{\mathrm{T}}_{\max }}{{\mathrm{T}}_{\mathrm{i}}}$×100.

Notes: ROAV_i denotes the ROAV of a volatile compound, while C_i and T_i denote the relative content (%) and threshold (μg/kg) of compound I, respectively. C_max and T_max represent the relative content (%) and threshold at which the compound contributes maximally to the aroma.

2.8. Statistical analysis

Data compilation was conducted utilizing Excel software. Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA) and permutation testing were carried out using SIMCA-P 14.1. One-way ANOVA with LSD post-hoc testing was conducted using SPSS 23.0. The results were presented as the mean of triplicate measurements (mean ± standard deviation). Radar plots, heatmaps, bar charts, and line graphs were produced utilizing GraphPad Prism 9.0.2 and Origin, whereas the heatmap was generated with TBtools. Metabolite chemical structures were generated utilizing ChemDraw version 22.0.0.

3. Results

3.1. Comparison of sensory profiles of TT, WT, LS, and HS

The sensory characteristics of white tea are significantly influenced by its shape during the processing of tea (Feng, Yang, et al., 2024; Zhou et al., 2023). Three distinct compressing methods yielded different flavor profiles when compared to TT (Fig. 1B and C, Table S1 and S2). TT exhibits a pronounced grassy aroma, which was notably reduced in compressed white tea (Fig. 1B). Among the compressed samples, HS displayed the weakest grassy aroma but the strongest sweet fragrance, while LS showed a moderately sweet aroma. In contrast, TT's sweetness was relatively subdued. WT exhibited greater pekoe aroma, floral and fruity aroma intensity, and aroma persistence in comparison to LS and HS, nearing the contents of TT. WT displayed greater sweet, fresh, and pekoe flavors compared to the other samples, nearing the profile of TT (Fig. 1C). WT demonstrated the lowest bitterness, while HS exhibited the highest astringency. HS exhibited greater thickness and bitterness compared to TT, WT, and LS, whereas TT demonstrated the lowest thickness and WT the least bitterness. The results indicated that compressing traditional white tea enhances its thickness, while steaming increases bitterness and reduces grassy notes. Compressing withered leaves into tea cakes resulted in a reduction of bitterness, an increase in sweetness and thickness, and the development of floral and fruity aromas.

The volume of white tea cake served as a critical criterion for assessing the morphology of tea leaves, which directly influences the aesthetic appeal, storage quality, and transportation costs associated with white tea. TT displayed a naturally loose structure, reaching the greatest volume of 71.25 ± 1.15 cm³. Following compression treatment, WT and LS demonstrated comparable volumetric reductions to 47 ± 1.59 cm³ and 46.5 ± 1 cm³, respectively, whereas HS was the smallest volume of 13.75 ± 0.23 cm³ (Fig. 1D). Despite having the smallest volume, HS exhibited a flavor profile after steaming that differed significantly from loose tea. WT, however, diminishes cake volume while effectively maintaining pekoe aroma and freshness.

3.2. Non-volatile metabolite profiling of TT, WT, LS, and HS

3.2.1. Catechin components profiling of TT, WT, LS, and HS

Catechins are essential flavonoids found in tea, and their contents affect both the concentration and bitterness of white tea (Ma, Ma, et al., 2022). The total polyphenol contents of TT, WT, LS, and HS were 111.01 ± 18.29 mg/g, 132.68 ± 5.80 mg/g, 106.00 ± 4.00 mg/g, and 117.22 ± 10.88 mg/g, respectively, with WT showing a significantly higher total catechin content than TT, LS, and HS (P < 0.05) (Fig. 2C-a). All tea samples contained nine identified catechin monomers. EGCG constituted the majority of catechins, accounting for approximately 61 %–64 % of the total, and plays a significant role in the taste quality of tea. Cluster analysis had identified LS and HS as a group, whereas WT was distinctly clustered apart from the other three samples. Within WT, the concentrations of EC, EGC, ECG, EGCG, and EGCG-CH₃ were significantly higher compared to TT, LS, and HS (P < 0.05) (Fig. 2A). This indicates that the compression of withered leaves significantly boosts the retention of these catechin monomers.

Fig. 2.

(A) The heatmap illustrates the catechin content of tea samples. (B) The heatmap illustrates the amino acid content of tea samples. (C) The content change of different metabolites. (D) The heatmap illustrates the sugar content of tea samples. The data are expressed as means ± standard deviation (n = 3), with significant differences indicated by different lowercase letters based on ANOVA, followed by the Duncan test, at P < 0.05. ‘H' indicates the high value of relative abundance, ‘L' indicates the low value of relative abundance.

The study examined the effects of compression methods and steaming treatment on the bitterness of white tea cakes. Catechins were categorized according to geometric isomerism and gallate moieties. Nine catechin monomers were classified into non-ester catechins (C, EC, GC, EGC) and ester catechins (CG, ECG, GCG, EGCG, EGCG-CH₃) (Shan et al., 2024). Among the four tea samples, the contents of ester catechins were substantially higher compared to non-ester catechins. Moreover, WT demonstrated a significantly higher content of ester catechins relative to TT, LS, and HS (Fig. 2C-b and c), suggesting that the compression method influences catechin contents in white tea cakes. The contents of ECG, EGC, and EGCG-CH₃, WT exhibited the highest contents. The lowest contents were observed in LS, highlighting a notable difference compared to WT (P < 0.05). Non-ester catechins GC and GCG exhibited significantly elevated contents in LS and HS relative to WT. WT enhanced the retention of catechin monomers, thereby increasing the richness of mellow and thick flavor in white tea.

3.2.2. Amino acid composition profiling of TT, WT, LS, and HS

Amino acids are essential compounds that influence the umami flavor of white tea while mitigating its bitter and astringent characteristics (Ma, Wang, et al., 2022). Amino acids play a vital role in tea processing by contributing to the formation of tea aroma, with volatile aldehydes and other transformed products serving as key components of this aroma. Some amino acids exhibited unique fragrances that were closely associated with the taste and aroma of tea, which is a critical factor in determining tea quality. Analysis of four samples resulted in the identification of 22 amino acids. The total amino acid content in TT, WT, LS, and HS were 10.35 ± 1.57 mg/g, 13.9 ± 0.23 mg/g, 12.27 ± 1.69 mg/g, and 12.57 ± 0.29 mg/g, respectively, with WT showing a significantly higher total amino acid content than TT (Fig. 2C-d). Cluster analysis demonstrated that WT and LS clustered together, whereas TT formed a separate cluster with the other three samples. This suggested that amino acid contents in white tea cakes were affected by compression intensity, highlighting differences among these samples (Fig. 2B). WT exhibited significantly elevated contents of histidine (His), lysine (Lys), theanine (Thea), and total amino acids in comparison to TT, LS, and HS (P < 0.05), suggesting that WT promoted the accumulation of amino acids in white tea cakes. The lysine (Lys) content in LS was significantly lower than in WT (P < 0.05), while the isoleucine (Ile) content in HS was significantly higher than in LS (P < 0.05). The findings suggest that the amino acid content in the white tea cake following steam pressing was influenced by the pressure intensity applied during the process.

The 22 free amino acids can be grouped into four categories based on their taste profiles: fresh (Glu, Gln, Asp, Asn, Thea), sweet (Gly, Ala, Trp, Met, Cys, Thr, Pro, Ser), bitter (Tyr, Arg, Phe, His, Lys, Ile, Leu, Val), and aromatic (Val, Leu, Phe, Arg, Lys). The predominant presence of fresh amino acids was noted, while sweet and bitter amino acids assume secondary roles (Feng, Zhuang, et al., 2024). Glutamate, theanine, and aspartate were significant contributors to umami, which was essential for the unique flavor profile of white tea cakes. WT demonstrated higher contents of total amino acids, aromatic amino acids, bitter amino acids, and fresh amino acids in comparison to TT, LS, and HS (Fig. 2C-e, f, h, and i). HS exhibited the highest content of sweet amino acids, followed by WT, which may account for HS increased sweetness. TT and LS exhibited comparable amino acid content preferences. The results indicated that compressing withered leaves into tea cakes increases total amino acid content, with theanine significantly contributing to the distinctive flavor of tea soup.

3.2.3. Sugar components profiling of TT, WT, LS, and HS

Sugar compounds play a crucial role in contributing to the sweet flavor and aroma formation of white tea (Wu et al., 2019). In four tea samples, nine soluble sugars were identified: five monosaccharides (glucose, fructose, galactose, ribitol, and fucose), two disaccharides (sucrose and lactose), and two trisaccharides (melezitose and raffinose). Fructose was the most abundant, followed by sucrose, while raffinose exhibited the lowest content. Cluster analysis categorized TT and WT as a group, while LS and HS formed another group. The contents of raffinose, melibiose, fructose, ribitol, glucose, and galactose in TT and WT exceeded those in LS and HS (Fig. 2C). Similarly, the contents of lactose and sucrose in HS exceeded those in TT and WT. The results demonstrated that the content of soluble sugars in white tea cake varies with different compression methods. Notably, the content was comparable between white tea cake compressed with withered leaves and TT, while significant differences exist among the various compression techniques.

Compared to TT, the content of soluble sugars in tea leaves decreased after compression, with the WT exhibiting a total sugar content of 24.14 ± 4.01 mg/g, surpassing that of both LS and HS (Fig. 2C-g). The variations in sugar compounds among white tea cakes compressed through different methods were mainly evident in the fructose content. WT exhibited a fructose content of 12.06 ± 2.42 mg/g, surpassing LS by 0.78 mg/g and HS by 1.02 mg/g, respectively. The results demonstrated that WT exhibited the highest content of soluble sugar preservation among the three compression methods analyzed. The conversion of soluble sugars took place in LS and HS due to hygrothermal action, with sugars contributing to the formation of aroma compounds, resulting in a reduction of their soluble sugar content. It was clear that compressing withered leaves into tea cakes facilitates the accumulation of sweet-taste compounds, while LS and HS enhanced the conversion of sugar compounds.

3.3. Volatile metabolites profiling of TT, WT, LS, and HS

3.3.1. Qualitative and quantitative analysis of volatile compounds

Conducting both qualitative and quantitative assessments of volatile compounds is essential for evaluating the overall quality of white tea (Chen et al., 2019; Wu et al., 2022). The unique aroma profile of white tea is intrinsically tied to its endogenous volatile compound synthesis, where the proportions of various volatile components shape its special characteristics. In the analysis of all tea samples, a total of 74 volatile compounds were identified, categorized into 13 alcohols, 19 esters, 11 ketones, 10 aldehydes, 7 acids, 9 hydrocarbons, and 5 other heterocycles (Fig. 3A). In the four tea samples, the primary volatile constituents were alcohols, which were notable for their floral, sweet, and fruity olfactory profiles. The relative alcohol content in WT was 59.72 % across 12 types, whereas in HS, it was 58.17 % across 13 types. Ester compounds contributing floral and fruity aromas to tea exhibited the highest relative content of 13.24 % (19 types) in WT and the lowest relative content of 9.43 % (17 types) in LS. Ketone compounds, arising mainly from processes such as oxidation, the Maillard reaction, and amino acid catabolism, contribute to the fruity and fresh aroma profiles. They exhibited higher relative contents of 5.56 % (11 compounds) and 6.6 % (11 compounds) in LS and HS, respectively, while TT displayed a lower relative content of ketones at 3.85 % (10 compounds). Among hydrocarbons, LS exhibited the highest relative content at 20.42 % (65 compounds), whereas WT demonstrated the lowest at 4.27 % (11 compounds) (Fig. 3B). The relative content of compressing after steaming into tea cakes increased, while that of compressing withered leaves into tea cakes decreased. This may be the reason why WT retains its freshness and fruity aroma.

Fig. 3.

(A) The proportions of volatile components across four tea samples. (B) A variety of distinct volatile substance categories. (C) The scatter plots depict OPLS-DA results for all tea samples. (D) OPLS-DA loading scatter plot. (E) Confirmation of the OPLS-DA model validity for the entire sample. (F) The variable importance in projection (VIP) scores for the tea samples, with the red column emphasizing volatile components scoring above VIP > 1. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.3.2. Analysis of key aroma compounds

Using OPLS-DA, 74 volatile compounds from four samples were analyzed, which clearly showed that different treatments were significantly separated in different image regions (Fig. 3C). Based on the VIP > 1 criterion (Fig. 3F), a total of 28 volatile components were identified, comprising 5 alcohols, 6 esters, 2 aldehydes, 5 ketones, 2 acids, 5 hydrocarbons, and 3 other compounds. These compounds could serve as biomarkers for differentiating different white tea cake samples. This loading scatter plot visualizes the distribution of compounds based on R2X [1] = 0.531 and R2X [2] = 0.209 (Fig. 3D), illustrating the relationship between X-variables (compounds) and the Y-variable (compression methods). X-variables proximal to the virtual Y-variable exhibited the highest intergroup discriminatory power. The plot further explains the differences in specific volatile components among the four white tea samples. The reliability of the OPLS-DA model was confirmed through 200 permutation tests (Fig. 3E).

To screen the characteristic volatile compounds of white tea cakes by different compressing methods, the ROAV of 28 volatile components was determined (Table 2). Based on the criterion of ROAV ≥0.1, nine key volatile compounds were screened, including (Z)-3-hexenyl isovalerate, nonanoic acid, 1-octen-3-ol, (E,E)-3,5-octadien-2-one, 4-oxoisophorone, nerolidol, 2, 6, 6-trimethylcyclohexanone, and hexanal, all of which exhibited ROAV ≥1. These compounds were crucial for providing floral, fruity, minty, sweet, and cheesy characteristics to white tea. Heptanal, with an ROAV range of 1 to 0.1, contributed herbaceous woody. Cluster analysis of the 9 key volatile compounds (Fig. 4A) indicated that TT and WT clustered together, whereas HS formed a separate cluster with the other three samples. In WT, nonanoic acid exhibited significantly higher contents compared to TT, LS, and HS. Conversely, (Z)-3-hexenyl isovalerate, 1-Octen-3-ol, (E,E)-3,5-octadien-2-one, and heptanal exhibited significantly higher contents compared to TT. In WT and TT, (Z)-3-hexenyl isovalerate, 1-octen-3-ol, (E,E)-3,5-octadien-2-one, 4-oxoisophorone, nerolidol, 2,6,6-trimethylcyclohexanone, hexanal, and heptanal demonstrated significantly lower contents in comparison to LS and HS. Nonanoic acid (cheese) was identified as the characteristic volatile compound of WT. (Z)-3-Hexenyl isovalerate (fruity, grassy), nerolidol (mild floral, sweet, woody), and 2,6,6-trimethylcyclohexanone (honey, pungent, cistus) were identified as characteristic compounds of LS. In contrast, (E,E)-3,5-octadien-2-one (fruity, grassy), 1-octen-3-ol (orange floral, mushroom), 4-oxoisophorone (Honey, rose), 1-octanol (grassy, green), and hexanal (grassy) were characteristic of HS. The results indicate that the aroma of flowers and fruits from compressing withered leaves into tea cakes was more pronounced, resembling that of traditional loose white tea. Hygrothermal action facilitated the volatilization of low-boiling-point aromatic compounds, altering the aroma profile of white tea and reducing grassy notes.

Table 2.

ROAV of key volatile components in different white tea cake pressing methods.

Compounds	CAS	RI	VIP	Odor description	OT	ROAV
	Number	Calculate			(μg/kg)	TT	WT	LS	HS
(Z)-3-Hexenyl hexanoate	31,501–11-8	1379.20	1.33	Fruity	781	0.00	0.00	0.01	0.00
(Z)-3-Hexenyl isovalerate	35,154–45-1	1226.74	1.24	Fruity, grassy	0.01	100	100	100	81.7
Heptanal	111–71-7	902.30	1.22	Herbaceous woody	2.80	0.76	0.37	0.26	0.70
Nonanoic acid	112–05-0	1292.43	1.22	Cheese	1.50	0.30	26.3	3.69	0.17
Octanoic acid	124–07-2	1197.91	1.21	Nasty smell	350	0.18	0.04	0.00	0.00
α-Cubebene	17,699–14-8	1342.60	1.17	Herbaceous woody	–	–	–	–	–
Methyl geranate	2349-14-6	1316.45	1.17	–	–	–	–	–	–
1-Octen-3-ol	3391-86-4	981.69	1.15	Orange floral, mushroom	1.00	5.06	6.53	4.60	14.2
(E,E)-3,5-Octadien-2-one	30,086–02-3	1094.61	1.14	Fruity, grassy	0.50	9.36	12.1	6.70	14.7
2-Heptanone	110–43-0	888.60	1.13	Fruity, sweet, herbaceous woody	600	0.02	0.02	0.01	0.03
4-Oxoisophorone	1125-21-9	1140.55	1.13	Honey, rose	0.05	69.2	50.0	42.4	100
6-Methyl-6-hepten-2-one	10,408–15-8	984.86	1.13	–	–	–	–	–	–
1-Octanol	111–87-5	1075.80	1.13	Grassy, green	2.70	0.00	0.00	0.00	2.67
β-Calacorene	50,277–34-4	1535.50	1.12	–	–	–	–	–	–
1-Pentanol	71–41-0	782.78	1.11	Pleasant odor	100	0.03	0.03	0.04	0.21
Nerolidol	7212-44-4	1559.40	1.11	Mild floral, sweet, woody	10.0	2.04	1.60	1.79	2.00
2-Butylfuran	4466-24-4	889.36	1.10	Fruity, sweet, spicy	–	–	–	–	–
(Z)-Caryophyllene	118–65-0	1411.53	1.10	Woody, spicy	–	–	–	–	–
(Z)-2-Penten-1-ol	1576-95-0	784.60	1.09	Fruity	720	0.00	0.00	0.00	0.02
5-Ethyl-6-methyl-3(E)-hepten-2-one	57,283–79-1	1138.34	1.07	Fruity, grassy	–	–	–	–	–
2,6,6-Trimethylcyclohexanone	2408-37-9	1032.85	1.07	Honey, pungent, cistus	0.10	20.3	11.1	34.5	25.8
δ-Octalactone	698–76-0	1282.52	1.06	Coconut	420	0.01	0.01	0.01	0.01
Hexanal	66–25-1	802.07	1.06	Grassy	5.00	2.00	1.19	0.86	1.88
2-Butyl-2-octenal	13,019–16-4	1369.69	1.06	Fruity	–	–	–	–	–
Methyl α-linolenate	301–00-8	2094.55	1.06	–	–	–	–	–	–
Methyl enanthate	106–73-0	1024.00	1.04	Floral, fruity, sweet	–	–	–	–	–
1-Ethylpyrrole	617–92-5	810.49	1.04	Charre	10.0	0.01	0.01	0.02	0.04
(E)-2-pentenylfuran	70,424–14-5	998.34	1.04	–	–	–	–	–	–

Note: Sorted by VIP value in descending order. CAS stands for Chemical Abstracts Service Registry Number. ‘0.00’ represents 0.01. ‘-’ means data was not found in the literature. OT signifies Odor Threshold (in water), and all OT values were sourced from the following references (Chen et al., 2023; Hao et al., 2023; Li, Li, et al., 2023; Ma et al., 2024; Tang et al., 2023.)

Fig. 4.

(A) The heatmap illustrates the differentially accumulated volatile compounds in tea samples with VIP > 1 and ROAV >0.1. (B) Correlation diagram of key aroma actives and aroma sensory properties. (C) Correlation diagram of key volatile compounds and taste sensory properties. Notes: orange represents positive correlation, purple represents negative correlation, and the larger the circle, the stronger the correlation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

4. Discussion

4.1. WT promotes the formation of fresh and sweet flavor

The flavor serves as a significant criterion for assessing the quality of white tea cakes. This study employed three compression methods: WT, LS, and HS. The flavor and aroma of white tea cakes are directly affected by various compression methods. Amino acids are metabolites that correlate positively with tea quality and contribute to the fresh flavor of white tea. Catechins contribute to the intensity and bitterness of white tea, whereas soluble sugars augment its sweetness. The interaction of these substances collectively contributes to the overall flavor profile of white tea (Chen, Shi, et al., 2020). Further transformations occurred in amino acids, catechins, and soluble sugars following the compression of white tea cake. Our observations indicate that the total content of free amino acids, soluble sugars, and catechins in the compressed white tea cake after steaming was lower than that of tea cakes made from compressing withered leaves. This suggests that hygrothermal action influences the quality formation of white tea cakes. The findings align with the results reported by Xiang et al., 2024, wherein moist-heat compression processing significantly elevated the concentrations of tea polyphenols, catechins, soluble proteins, flavonoids, and gallic acid, thereby enhancing the compositional richness of tea infusions. Conversely, amino acids and soluble sugars were notably reduced, potentially due to Maillard reactions induced by hydrothermal treatment. Research indicates that hygrothermal action during the steaming of tea relies on heat conduction and hydrodynamics, serving as the primary driving force after enzymatic action. This process facilitates the non-enzymatic oxidation and decomposition of the components involved (Li et al., 2022; Zhang et al., 2017). For the same loose white tea, WT, LS, and HS exhibited higher amino acid content compared to TT, with WT showing the highest concentrations of free amino acids. Amino acids in white tea primarily originate from protein hydrolysis during withering. Previous research has shown that enzymes like proteases gradually break down proteins into free amino acids (e.g., GABA, aspartic acid, lysine acid) under mild conditions (Tian et al., 2024; Zhou et al., 2023). While enzymatic proteolysis dominates, dehydration stress during withering may also indirectly promote amino acid accumulation. Previous scholars proposed that dehydration activates stress-responsive genes, potentially enhancing amino acid mobilization. Additionally, Hygrothermal action (e.g., steam treatment in cake compression) halts hydrolysis and even degrades existing amino acids, while WT maintains hydrolysis activity, leading to accumulation (Zhu et al., 2015). However, non-enzymatic reactions like Maillard reactions dominate under high heat, converting amino acids into aldehydes, ketones, or other aromatic volatiles. These reactions are suppressed in WT due to minimal mechanical disruption and no Hygrothermal action (Xie et al., 2023). The WT preserves cell integrity, limiting oxidative interactions between amino acids and polyphenols. Thus, preserving the freshness and sweetness of white tea loose tea. While our study did not directly measure enzyme activity, the consistency with proteomic and metabolomic literature strongly supports the above mechanisms. Future studies will quantify enzymes like proteases activity under different compression conditions. Track isotope-labeled amino acids to distinguish hydrolysis vs. synthesis contributions. This result aligns with previous studies suggesting a slight reduction in free amino acid content due to hygrothermal action. (Li et al., 2017).

Catechins significantly contribute to the taste of tea, with prevailing research indicating their essential role in imparting bitterness and astringency (Liu et al., 2023; Zhang et al., 2020). The findings of this study indicate that the catechin content increases following the compression of withered leaves, whereas Hygrothermal treatment does not significantly affect catechin contents in white tea. This can be attributed to the increase of polyphenolic compounds under mechanical force, while hygrothermal action results in the oxidation and degradation of these compounds. The total polyphenol and EGCG contents in WT were significantly higher than those in LS and HS after steaming (Table S4). This is attributed to non-enzymatic oxidation occurring under Hygrothermal action, which accelerates the oxidation of catechins to yield quinones. These quinones subsequently polymerize to form oxidized products such as theaflavins. Our findings align with previous studies by Fan et al. (2016), where moist-heat treatment was shown to markedly reduce catechin levels while promoting the generation of polyphenolic oxidation products. Catechins are highly sensitive to enzymatic and non-enzymatic oxidation. In black tea and Pu-erh tea, oxidation via polyphenol oxidase (PPO) significantly reduces catechin levels by converting them into theaflavins (TFs) and thearubigins (TRs) (Li et al., 2021; Wang et al., 2024; Zhu et al., 2015). However, white tea processing typically involves minimal fermentation, which partially preserves catechins (Zhou et al., 2023). In our study, the WT sample exhibited significantly lower oxidative reaction compared to other samples, likely owing to the absence of Hygrothermal action and lower mechanical force during its compression process (Zhou et al., 2023; Zou et al., 2022). Catechins can polymerize into high-molecular-weight compounds like theabrownins (TBs) during prolonged thermal processing (Wang et al., 2024). Studies suggest that mechanical compression (e.g., steaming or pressing) accelerates these reactions by disrupting cellular structures and enhancing enzyme-substrate interactions (Fan et al., 2023). However, the WT milder compression conditions might limit polymerization, preserving monomeric catechins. This aligns with previous studies showing that extended drying degrades catechins and elevates TBs, altering flavor profiles. The variation in catechin content may be attributed to the direct compression of withering leaves in the absence of hygrothermal action.

Soluble sugars amplify tea sweetness and function as aroma precursors. A decline was observed in the soluble sugar content of WT, LS, and HS, but the reduction in soluble sugar content in white tea cake after steaming was greater than that observed in tea cakes made from compressing withered leaves. Previous studies suggest that the contents of amino acids and soluble sugars initially increased due to proteolysis and the degradation of tea polysaccharides, but these contents subsequently decreased as a result of various thermochemical reactions, including Maillard and caramelization. Concurrently, glycosidase-induced hydrolysis of glycoside-linked volatile compounds takes place, enabling the generation of aroma components (Ntezimana et al., 2021). Prior studies (Peng et al., 2024) indicate that elevated temperatures promote the generation of Maillard-derived compounds, such as pyrazines, furans, and pyrroles, which are associated with roasted, caramel-like, or nutty flavor notes (Liu et al., 2022; Wang et al., 2022). These conditions also accelerate Maillard reaction kinetics. However, the relatively low abundance of reducing sugars and free amino acids in white tea cake after steaming may constrain the extent of Maillard reaction progression. This aligns with our volatile profile data, which revealed higher ROAV values for 1-ethylpyrrole (a Maillard reaction product) in LS and HS compared to WT.

To elucidate the correlation between non-volatile components and taste attributes, a correlation analysis was performed between key non-volatile components (P < 0.05) and QDA (Fig. 4C). The components exhibiting a positive correlation with umami, sweet, and pekoe taste were theanine, isoleucine, phenylalanine, lysine, EGC, EGCG-CH₃, ribitol, raffinose, melezitose, glucose, galactose, and theaflavin-3,3-gallate. In contrast, EGCG-CH₃, ribitol, raffinose, melezitose, glucose, galactose, and theaflavin-3,3-gallate exhibited negative correlations with bitterness and astringency. Phenylalanine, lysine, and EGC demonstrated negative correlations with astringency. GCG, CG, and GC exhibited a positive correlation with mellow and thick attributes while showing a negative correlation with umami, sweet, and pekoe tastes. Amino acids and soluble sugars significantly contributed to umami, sweet, and pekoe tastes, whereas GCG, CG, and GC in polyphenols primarily influenced mellow and thick flavors, while also contributing to bitterness and astringency. In alignment with prior research findings, GCG, CG, and GC classifications were categorized as trans catechins, which were characterized by bitterness and astringency. Phenylalanine, ribitol, raffinose, melezitose, glucose, and galactose were significant non-volatile components influencing the sensory flavor of white tea cake, with raffinose exhibited the strongest correlation to umami, sweetness, and pekoe (Shan et al., 2024).

In summary, the content of catechins and amino acids in the compressed withered leaves of tea cakes significantly increased, enhancing the mellow, thick, and fresh qualities of the tea flavor. However, a reduction in soluble sugars was observed, potentially attributable to subsequent thermal chemical reactions, including the Maillard reaction. The flavor composition of traditional white tea cake, pressed following hygrothermal action, significantly influences the overall flavor profile, while the effect on the flavor composition of compressed withered leaves in tea cakes was comparatively minimal.

4.2. WT is beneficial to the formation of floral and fruit aromas

The essence of tea quality is largely determined by its aroma. Among the three compressing methods, volatile compounds exhibit significant differences, including (Z)-3-hexenyl isovalerate, 1-octanol, nonanoic acid, 1-octen-3-ol, (E, E)-3,5-octadien-2-one, 4-oxoisophorone, nerolidol, 2, 6, 6-trimethylcyclohexanone, and hexanal. These compounds were likely indicative of the three types of white tea cakes. Previous studies on the flavor components of white tea indicate that the fresh and floral aroma characteristics were attributed to a high proportion of alcohol compounds (Wu et al., 2022). The varying compositions of these components contribute to the development of diverse flavors, including grass, sweet floral, and fruity notes (Chen, Zhu, et al., 2020). Nerolidol and hexanal were the two compounds with the highest relative content among those that exhibited significant differences, contributing to the fruit and sweet aroma of white tea. Nonanoic acid was associated with cheese flavor, 4-oxoisophorone presents a sweet and rose scent, while (E, E)-3,5-octadien-2-one was characterized by fruity and grassy notes, and was previously recognized as a contributor to the fresh aroma of Japanese green tea (Wu et al., 2022). Further analysis revealed significant differences in the relative content of volatiles among the three types of white tea cakes. In LS and HS, the relative concentrations of alcohols, acids, and esters were lower compared to WT, whereas the relative concentrations of ketones, aldehydes, and hydrocarbons were higher than WT. In LS and HS, compared with WT, the relative content of alcohols was reduced by 1.55 % and 7.17 %, the relative content of esters was lowered by 3.81 % and 3.71 %, and the relative content of acids was decreased by 10.88 % and 3.94 %. Conversely, the relative content of ketones increased by 1.29 % and 2.33 %, aldehydes exhibited a rise of 0.57 % and 0.76 %, and hydrocarbons exhibited a rise of 13.53 % and 10.92 %. This observation is consistent with our preliminary studies, which demonstrated that post-steaming compression reduces alcohol derivatives while increasing the average relative content of alkenes, ketones, and hetero‑oxygenated compounds. The observations indicate that white tea cakes made from withered leaves may preserve a greater amount of the floral and fruity aromas traditionally associated with white tea.

To investigate the relationship between the identified key volatile compounds and aroma characteristics, a correlation analysis was performed, focusing on 9 key volatile components (VIP > 1, ROAV >0.1) and the QDA index score (Fig. 4B). Six key volatile compounds were identified: 1-octanol, 4-oxoisophorone, 1-octen-3-ol, nerolidol, 2, 6, 6-trimethylcyclohexanone, and hexanal. These compounds exhibited a positive correlation with sweet aroma, while showing a negative correlation with aroma strong, floral-fruity aroma, pekoe aroma, grassy aroma, and lasting fragrance. Two key volatile compounds—(Z)-3-hexenyl isovalerate and (E, E)-3,5-octadien-2-one—exhibited a positive correlation with aroma fragrant and lasting, and sweet aroma. Conversely, these compounds showed a negative correlation with strong, floral-fruity, pekoe, and grassy aromas. Nonanoic acid primarily contributed to floral and fruity aromas, pekoe aroma, and a fragrant, lasting quality, while (Z)-3-hexenyl isovalerate, 1-octanol, 1-octen-3-ol, (E, E)-3,5-octadien-2-one, 4-oxoisophorone, nerolidol, 2, 6, 6-trimethylcyclohexanone, and hexanal contributed to sweet aroma. The findings suggest that (Z)-3-hexenyl isovalerate, 1-octanol, nonanoic acid, 1-octen-3-ol, (E, E)-3,5-octadien-2-one, 4-oxoisophorone, nerolidol, 2, 6, 6-trimethylcyclohexanone, and hexanal were significant volatile components influencing the sensory flavor of white tea cake. Notably, nonanoic acid exhibited the strongest correlation with robust floral and fruity aromas, fragrant and lasting, as well as pekoe aroma. 4-Oxoisophorone exhibited the strongest correlation with sweet aroma.

In conclusion, different compression methods markedly affect the volatile composition of white tea cakes. The relative content of alcohols, acids, and esters declined. Conversely, the relative content of ketones and hydrocarbons increased, while aldehydes exhibited a slight increase with minimal change. The nerolidol in alcoholic substances was associated with floral and sweet aromas, whereas (E, E)-3,5-octadien-2-one was linked to fruity aromas. We have verified multiple batches of raw materials in production, and the test results show that the quality change trend is consistent. In the future, further research can be conducted on the effects of differences among varieties in raw materials on quality and content conversion. In addition, the influence of moisture content variations in withered tea leaves on white tea cake compression quality will be a key research focus for our team in subsequent studies.

5. Conclusions

This study conducted a metabolomic analysis of three different compressing methods to examine their effects on non-volatile and volatile compounds in white tea cakes. The compression of withered leaves into tea cakes led to a reduction in soluble sugars and an increase in catechins and amino acids. The study identified significant alterations in the aroma compounds resulting from different cake compressing methods, particularly noting an increase in alcohol, acid, and ester substances that contributed to the enhancement of floral and fruity characteristics. In conclusion, the technique of compressing withered leaves preserves more of the flavor of loose white tea, enhancing both the freshness and sweetness of the white tea cake while also developing a pleasant floral and fruity flavor. These findings enrich the understanding of white tea cake processing and provide valuable insights for quality control and assurance in white tea production.

CRediT authorship contribution statement

Hongzheng Lin: Writing – review & editing, Project administration, Methodology, Data curation. Shuping Ye: Writing – original draft, Investigation, Formal analysis, Data curation. Jiao Feng: Investigation, Formal analysis. Jinyuan Wang: Investigation, Data curation. Weiyi Kong: Writing – review & editing, Methodology. Junyang Wu: Writing – review & editing. Fangting Zhang: Writing – review & editing. Jiake Zhao: Writing – review & editing. Jiayi Guo: Writing – review & editing. Kaiyang Chen: Resources. Bugui Yu: Resources. Yun Sun: Writing – review & editing. Zhilong Hao: Writing – review & editing, Project administration, Methodology, Investigation.

Ethics statement

The authors ensure that the work described has been conducted in compliance with The Code of Ethics of the World Medical Association (Declaration of Helsinki) for experiments involving humans.

The ethical approval of sensory evaluation is not required by national laws. No human ethics committee or formal documentation process is available for sensory evaluation.

The authors affirm that appropriate protocols have been used to protect the rights and privacy of all participants. This includes ensuring that participation is voluntary, providing full disclosure of study requirements and risks, obtaining verbal consent from participants, not disclosing participant data without their knowledge, and allowing participants to withdraw from the study at any time.

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 1

Supplementary material 2

Supplementary material 3

Data availability

Data will be made available on request.