Mitigating long-term deterioration of frozen dough: Isothermal-buffered frozen storage for stabilizing water mobility and the gluten-starch network Authors: Shuang Li, Zixuan Wu, Xianghong Li, Yiqun Huang, Yongle Liu, Faxiang Wang Published: 2026 ABSTRACT Long-term quality deterioration remains a major challenge for the frozen dough industry. This study evaluated an isothermal-buffered frozen storage (IFS) against conventional air frozen storage (AFS) over 60 days. Results showed that the high thermal storage capacity of IFS acted as a physical barrier, reducing temperature fluctuations by 2.7-fold compared to AFS. This thermal stability effectively inhibited ice recrystallization, minimizing mechanical stress on the gluten-starch network and yeast cells. Consequently, IFS delayed the migration of bound water to a freezable state, preserving dough hydration. Bread from IFS-treated dough exhibited 15.69% higher specific volume and superior crumb uniformity. Mechanistically, the thermal buffering effect of IFS mitigates environmental temperature fluctuations, thereby limiting ice recrystallization and reducing subsequent mechanical damage to the gluten-starch matrix and yeast cells. This study offers a feasible strategy to enhance the stability of frozen dough against the unavoidable temperature variations commonly encountered in cold chain logistics. Introduction Frozen dough technology is essential to the modern baking industry, allowing for standardized production and fresh-on-demand availability (Zhang, Guo, et al., 2024). However, long-term frozen storage typically leads to quality deterioration, such as reduced bread volume and harder texture. Currently, air frozen storage (AFS) is the standard method in cold chain logistics; yet its low thermal conductivity makes it highly susceptible to temperature fluctuations during operations like transport or defrost cycles (Liu, Gu, et al., 2024). Thermodynamically, these fluctuations drive the continuous melting and refreezing of ice crystals (recrystallization) (Zhang et al., 2020). The resulting large, irregular ice crystals mechanically puncture the gluten network, weakening the non-covalent interactions that bind water molecules to structural proteins. Consequently, tightly bound water migrates into a freezable state, further accelerating ice crystal growth and compromising yeast viability (Omedi et al., 2019). While this vicious cycle of water migration and network disruption is a recognized challenge, practical strategies to fundamentally shield the complex dough matrix from external thermal shocks remain limited. Therefore, exploring alternative storage environments with higher thermal inertia is imperative. In contrast, isothermal-buffered frozen storage (IFS) offers a promising alternative by utilizing a medium with high heat capacity. This medium serves as a thermal buffer, effectively dampening the impact of external temperature fluctuations and maintaining a more stable environment for the samples (Cui et al., 2025). While previous studies have shown that IFS can slow down quality loss in muscle tissues by minimizing temperature variation, its effectiveness in protecting frozen dough, a far more complex system, remains to be fully understood during long-term storage. Unlike muscle tissue, dough consists of an intricate gluten-starch network and living yeast cells, which respond differently to environmental stress. Most existing research on IFS has focused on animal-based products or initial freezing rates (Liu, Li, et al., 2024; Zhang et al., 2022; Zhou et al., 2025). However, the primary challenge in the baking industry is sustaining microscopic integrity over extended periods, such as 60 days. Understanding how a stable storage environment influences the dough's microstructure is essential. This study aims to provide a practical entry point for the industry, showing how a buffered thermal environment can safeguard the dough's structure and fermentation capacity under the less-than-ideal conditions often found in real-world logistics. This study evaluated the protective effects of IFS on frozen dough over a 60-day period, using conventional AFS as a comparison. We employed thermodynamic analysis and LF-NMR to track water migration, alongside microscopic imaging to examine the structural integrity of the gluten network. By linking the stability of the storage environment and the inhibition of ice recrystallization to the final baking quality, we aim to clarify how a buffered thermal environment helps sustain dough longevity. Rather than proposing a final solution, these findings are intended to provide a practical reference for improving cold chain practices, illustrating how managing temperature stability can safeguard product quality in challenging real-world logistics. Materials and methods Materials Wheat flour, containing 11.1% protein, 1.5% fat, and 75.3% carbohydrates (wet basis), was supplied by Kemen Noodle Manufacturing Co., Ltd. (Changsha, China). Ingredients also included commercial dry yeast from Angel Yeast Co., Ltd. (Hubei, China), with all supplementary chemicals being of analytical grade. Sample preparation and frozen storage Dough was prepared by mixing wheat flour (200 g), salt (2 g), sucrose (10 g), high-activity dry yeast (0.8 g), and distilled water (100 mL) using an automatic kneading device for 10 min to achieve a smooth, homogeneous texture. The dough was then divided into 30 g portions, placed into 4 × 4 × 4 cm molds and manually flattened to a height of 2 cm, then the molds containing the dough were individually vacuum-sealed in 10 × 15 cm high-barrier aluminum foil pouches at a pressure of approximately −0.09 MPa. This combination of the internal mold and the external hermetic vacuum seal provided a reliable double physical barrier. It effectively eliminated the risk of any direct or indirect contact between the food product and the CaCl₂ coolant, thereby strictly preventing solvent penetration and securing food safety during the immersion process. The samples were randomly assigned to two groups and stored in a freezer (BC/BD-428HD, Haier Co., Ltd., Qingdao, China). Subsequently, the AFS group, with identical molds and vacuum packaging, was placed directly on the freezer shelves, while the IFS group was fully submerged in a pre-cooled (−20 °C) 280 g/L CaCl₂ solution housed within the same freezer. This method utilizes the high heat capacity of the liquid medium to dampen temperature oscillations during storage. Both groups were maintained at −20 ± 2 °C for 0, 1, 7, 15, 30, and 60 days. Before analysis, all samples were gently thawed at 4 °C for 12 h to ensure uniformity. Measurement of core temperature profiles A K-type thermocouple probe was inserted into the geometric center of the dough sample. Temperature data were recorded at 30 s intervals. The collected data were used to plot the freezing curves and to characterize temperature fluctuations during the frozen storage period. Determination of proofing time for frozen dough The proofing time was determined according to the method described by Öhgren et al. (2016), with minor modifications. After frozen storage, the dough samples were thawed at 4 °C for 6 h. Subsequently, the samples were transferred to a proofing cabinet maintained at 30 °C and 85% relative humidity (RH). The time required for the top of the dough to contact the sealing film was recorded as the proofing time. Deternation of fermentation capacity The fermentation capacity of the frozen dough was evaluated using the liquid displacement method, in accordance with the National Standard GB/T 20886–2007. Briefly, 50 g of dough was placed into Bottle A. Bottle B contained 250 mL of a displacement solution (prepared by dissolving 100 g NaCl and 10 mL concentrated H₂SO₄ in water to a final volume of 1 L). After assembling the system and verifying its airtightness, the apparatus was maintained in a constant temperature water bath at 30 °C. The CO₂ production was quantified by monitoring the volume of liquid displaced into Bottle C over a 3 h period (mL). Preparation of bread The bread-making procedure was conducted following the protocols previously established by Cui et al. (2021). Thawed dough samples were proofed in a fermentation cabinet for 150 min. Subsequently, they were baked in an oven (preheated to 180 °C) for 18 min. The resulting bread was allowed to cool for 0.5 h prior to further analysis. Determination of specific volume The specific volume of the bread was determined using the rapeseed displacement method, following the procedure described by Ramli et al. (2025). The specific volume was calculated using Eq. (1), where C represents the specific volume (mL/g), V is the volume of the bread (mL), and M is the mass of the sample (g).(1)C=VM Analysis of bread pore structure Images of the bread crumb (3 × 3 cm) were captured from the central region of the sample. The pore distribution characteristics were analyzed using ImageJ software (National Institutes of Health, USA), following the procedure described by Cui et al. (2023). Briefly, the images were converted into 8-bit grayscale format and subsequently binarized. With the detection threshold set for a pore area range of 0.0001 to 100 mm2, the total number of pores and the average pore area were calculated. Analysis of water migration in frozen dough Water migration characteristics were evaluated using a Low-Field Nuclear Magnetic Resonance (LF-NMR) analyzer (Zhu et al., 2022). The prepared dough samples were placed into 60 mm diameter NMR tubes and inserted into the instrument probe, which was maintained at a constant temperature of 32 °C. Transverse relaxation measurements (T2) were performed using the Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence with the following acquisition parameters: a waiting time of 2000 ms, an echo time of 0.2 ms, a sampling frequency of 250 kHz, and a number of scans set to 8. The obtained T2 relaxation decay curves were subsequently processed using distributed exponential fitting analysis to characterize the water distribution and migration within the dough matrix. Determination of freezable water content The freezable water content (FW) of the frozen dough was evaluated using differential scanning calorimetry (DSC), following the method described by Cui et al. (2021). Approximately 10 mg of the dough sample was accurately weighed into a TA aluminum pan. The samples were stored at −20 °C for 48 h prior to immediate analysis. During the DSC run, the samples were equilibrated at −20 °C for 5 min and then heated to 15 °C at a scanning rate of 3 °C/min. The endothermic melting enthalpy (ΔHm J/g) was recorded from the thermograms, and the FW was calculated according to Eq. (2), where ΔHm represents the melting enthalpy of the sample (J/g), ΔfusHm represents the latent heat of fusion of pure water (334 J/g), W is the total water content of the sample (g/g).(2)Fw%=ΔHₘΔfusHm×W×100 Observation of ice crystal structure in frozen dough Small dough specimens (5 × 5 × 5 mm) were excised from the geometric center of the frozen dough and immediately transferred into Carnoy's fixative (ethanol: glacial acetic acid = 3:1, v/v), which had been pre-cooled to −20 °C. The samples were fixed for 24 h. Subsequently, the specimens underwent a graded ethanol dehydration series: they were immersed in 70%, 80%, and 90% ethanol for 1 h each, followed by two changes of 95% ethanol and 100% ethanol for 20 min each. After dehydration, the samples were cleared with xylene and embedded in paraffin wax. Sections were cut to a thickness of 7 μm using a microtome. Finally, the sections were dried at 37 °C, deparaffinized, stained with hematoxylin and eosin (H&E), and examined using an optical microscope to analyze the tissue microstructure. Characterization of protein network structure The protein network structure of the frozen dough was characterized using Confocal Laser Scanning Microscopy (CLSM), following the method described by Wang et al. (2015) with minor modifications. Dough slices were stained with a 0.1 mg/mL Rhodamine B solution for 1 min, rinsed gently with deionized water, and blotted dry with filter paper to remove surface moisture. The stained samples were observed using a CLSM system equipped with an Apo 20×/0.75 objective lens. The excitation wavelength was set to 534 nm, and the emission signals were collected using a 590/50 nm band-pass filter. The acquired images were subsequently analyzed using Angiotool software to quantify the network characteristics. Microstructure characterization The microstructure of the frozen dough was visualized using Scanning Electron Microscopy (SEM) according to the method described by Zhang, Zhang, et al. (2024). The frozen dough was cut into small specimens (5 × 5 × 5 mm), kept frozen for 24 h, and subsequently fixed in Carnoy's fixative at −20 °C for 24 h. The fixed samples were dehydrated through a graded ethanol series (70%, 80%, and 90% for 1 h each), followed by two changes each of 95% and 100% ethanol (20 min per change). The dehydrated specimens were then vacuum freeze-dried, sputter-coated with gold for 30 s, and examined under an SEM at a magnification of 1000 × . Statistical analysis All experiments were performed in at least triplicate, and the results are expressed as mean ± standard deviation. Figures were generated using Origin 2024 software (OriginLab Corp., Northampton, MA, USA). Statistical analyses were performed using IBM SPSS Statistics 26.0 (IBM Corp., Armonk, NY, USA). Significance testing was conducted using Welch's ANOVA (α = 0.05). For post hoc multiple comparisons, the Least Significant Difference (LSD) test was selected when variances were homogeneous, whereas Dunnett's T3 test was employed when variances were unequal. Different lowercase letters in the figures and tables indicate statistically significant differences between groups (p < 0.05). Results and discussion Freezing behavior and thermal stability during storage The freezing process in food systems involves three critical stages: pre-cooling, phase transition (generally −5 °C to −1 °C), and sub-cooling. As shown in Fig. 1(a), IFS demonstrated significantly higher heat transfer efficiency, traversing the zone of maximum ice crystal formation 5.5 times faster than AFS (26.5 min). While this rapid freezing is essential for establishing a favorable initial microstructure with small ice nuclei, the long-term integrity of the dough depends even more on thermal stability during storage. (Zennoune et al., 2022).Fig. 1Changes in central temperature of dough during freezing process (a) and frozen storage (b). AFS and IFS represent frozen dough subjected to air frozen storage and isothermal-buffered frozen storage, respectively. AFS: air frozen storage; IFS: isothermal-buffered frozen storage.Fig. 1 Changes in central temperature of dough during freezing process (a) and frozen storage (b). AFS and IFS represent frozen dough subjected to air frozen storage and isothermal-buffered frozen storage, respectively. AFS: air frozen storage; IFS: isothermal-buffered frozen storage. After reaching the target −20 °C, the thermal histories diverged. While AFS suffered from marked fluctuations (−17.00 °C to −21.45 °C) due to defrost cycles and low air conductivity, the IFS group leveraged the high heat capacity of its medium to act as a steady-state thermal buffer, effectively controlling the temperature fluctuations during storage process. This buffering effect not only reduced the fluctuation amplitude by 2.7-fold compared to AFS, but also dramatically dampened the dynamic thermal response, suppressing the fluctuation rates from approximately 0.04 °C/min in the AFS group to merely around 0.01 °C/min. This control over temperature fluctuations is critical because thermal instability is the primary driver of ice recrystallization (Zhang et al., 2020). In the AFS group, the frequent temperature rises allow smaller ice crystals to melt, while the subsequent drops cause this water to redeposit onto larger crystals. This continuous cycle leads to the progressive enlargement of ice crystals. Consequently, while IFS offers the advantage of rapid initial freezing, its ability to suppress temperature fluctuations and inhibit recrystallization is likely the dominant factor contributing to the long-term preservation of frozen dough quality (Zhao et al., 2022). Evolution of proofing time during frozen storage Proofing time is a direct reflection of yeast metabolic activity and gas production rate. As shown in Fig. 2a, the proofing duration for both groups progressively increased with storage time. However, a distinct divergence emerged by the end of the 60-day period: the proofing time for the AFS group extended to 236 min, whereas the IFS group required approximately 47 min less to reach the target height.Fig. 2Effects of IFS on (a) proofing time, (b) CO₂ production, and (c) bread specific volume. AFS: air frozen storage; IFS: isothermal-buffered frozen storage. Different uppercase and lowercase letters indicate significant differences between frozen methods and storage times, respectively (p < 0.05).Fig. 2 Effects of IFS on (a) proofing time, (b) CO₂ production, and (c) bread specific volume. AFS: air frozen storage; IFS: isothermal-buffered frozen storage. Different uppercase and lowercase letters indicate significant differences between frozen methods and storage times, respectively (p < 0.05). This delay in the AFS group is primarily attributed to the cumulative mechanical stress on yeast cells. Temperature fluctuations inherent to air-based storage promote ice recrystallization, leading to the formation of larger ice crystals that damage yeast cell membranes (Yu et al., 2020). This cryo-injury suppresses yeast metabolic activity, thereby slowing down CO₂ production. In contrast, IFS acts as a thermal buffer, maintaining a steady-state environment that minimizes recrystallization (Cui et al., 2023). By protecting yeast viability from thermal stress, IFS effectively mitigates the decline in fermentation rate associated with long-term frozen storage. Fermentation capacity and specific volume of bread The specific fermentation capacity of the dough and the specific volume of the resulting bread are critical indicators of baking quality, reflecting both gas production and gas retention capabilities. As illustrated in Fig. 2(b, c), both parameters declined over the 60-day storage period. However, IFS exhibited significantly better preservation. By the end of storage, the fermentation capacity of the IFS group (387 mL kg−1 h−1) was approximately 107 mL kg−1 h−1 higher than that of the AFS group. Consequently, the IFS bread maintained a specific volume of 1.72 mL/g, representing a 15.69% improvement over the AFS group (1.56 mL/g, p < 0.05). The deterioration observed in the AFS group is largely driven by the disruption of the gluten network, which occurs through two pathways exacerbated by the unstable thermal history (Zhang et al., 2020). Physically, the large and irregular ice crystals formed via recrystallization can mechanically damage the gluten matrix, directly compromising its continuity. Biochemically, cryo-injury to yeast cells (as noted in Section 3.2) triggers the release of endogenous reducing substances, such as glutathione. These substances can cleave the disulfide bonds within the gluten network, chemically weakening the dough structure (Ding et al., 2015; Lu et al., 2025). Macroscopically, the specific volume of bread is governed by the gas retention capacity of the dough, which inherently relies on this structural integrity. In the AFS group, the fragmented protein network exhibits a reduced capacity to act as a continuous viscoelastic film, leading to the escape of CO₂ during proofing and baking. Conversely, by restricting temperature fluctuations, the IFS strategy helps limit both the physical damage and the chemically induced weakening. This thermal buffering assists in preserving a more intact and continuous gluten structure. Functionally, this preserved network serves to better entrap the gas produced by surviving yeast cells, thereby providing a structural basis for the 15.69% improvement in specific volume. Visual characterization of bread crumb porosity The characteristics of gas cells, specifically their distribution and size, are key indicators of the gluten network's integrity (Han et al., 2026). As illustrated in Fig. 3(a, b), fresh bread typically displays a uniform and fine pore structure. However, following frozen storage, the crumb texture became noticeably coarser and denser. This structural deterioration is closely linked to the behavior of ice crystals during storage. In the AFS group, unavoidable temperature fluctuations tend to accelerate ice recrystallization. This process causes ice crystals to grow and change shape, physically puncturing the thin walls of the gas cells. As these walls rupture, adjacent small pores merge to form larger, irregular voids, reducing the dough's overall gas retention capacity (Meziani et al., 2012).Fig. 3Effects of IFS on the cross-section structure. (a) Cross-sectional view of bread; (b) gray-scale transition map of bread cross-section; (c) figure of quantitative analysis of stomatal density of bread cross-section; (d) figure of quantitative analysis of average stomatal area of bread cross-section. AFS: air frozen storage; IFS: isothermal-buffered frozen storage. A7, A15, A30, and A60 represent AFS samples stored for 7, 15, 30, and 60 days, respectively, while I7, I15, I30, and I60 represent IFS samples stored for the corresponding durations.Fig. 3 Effects of IFS on the cross-section structure. (a) Cross-sectional view of bread; (b) gray-scale transition map of bread cross-section; (c) figure of quantitative analysis of stomatal density of bread cross-section; (d) figure of quantitative analysis of average stomatal area of bread cross-section. AFS: air frozen storage; IFS: isothermal-buffered frozen storage. A7, A15, A30, and A60 represent AFS samples stored for 7, 15, 30, and 60 days, respectively, while I7, I15, I30, and I60 represent IFS samples stored for the corresponding durations. Quantitative analysis (Fig. 3c, d) confirms this observation. After 60 days, the IFS group exhibited a pore density approximately 11.47 cells/cm2 higher than that of the AFS group, alongside an average pore area that was 0.09 mm2 smaller. These differences suggest that the thermal stability provided by IFS plays a protective role. By acting as a thermal buffer, IFS limits the temperature fluctuations that drive the merging of ice crystals(Liu, Zhang, et al., 2024). Consequently, the microscopic architecture of the dough is better preserved, resulting in a finer and more uniform bread crumb compared to the coarser texture observed in the AFS group. Water distribution and migration properties The distribution and binding state of water are critical determinants of dough rheology and baking performance. Fig. 4(a) presents the typical T2 relaxation distribution, identifying three distinct water populations: tightly bound water (T21), immobilized water (T22), and free water (T23) (Gong et al., 2019). Over the storage period, a clear shift in water status was observed (Fig. 4b), characterized by a depletion of bound water and a corresponding rise in more mobile water fractions. This redistribution mirrors the stability of the dough's internal environment. In the AFS group, the unstable thermal history acts as a catalyst for water migration. Continuous temperature variations disturb the thermodynamic equilibrium between water molecules and the gluten-starch matrix. This thermal agitation likely weakens the specific binding sites on the protein surface, facilitating the detachment of water molecules and their transition into a free, mobile state (Jiang, Guo, & Zhu, 2025; Liu et al., 2023).Fig. 4Effects of IFS on the water status and distribution in frozen dough: (a) T2 relaxation time distribution curves, (b) peak area proportions of water populations (T21, T22, T23), and (c) freezable water content. AFS: air frozen storage (AFS); IFS: isothermal-buffered frozen storage.Fig. 4 Effects of IFS on the water status and distribution in frozen dough: (a) T2 relaxation time distribution curves, (b) peak area proportions of water populations (T21, T22, T23), and (c) freezable water content. AFS: air frozen storage (AFS); IFS: isothermal-buffered frozen storage. Quantitatively, the IFS group exhibited a much higher degree of stability. The reduction in bound water (T21) for IFS samples (1.28%) was nearly half that of the AFS samples (2.31%). Furthermore, the increase in free water (T23) in the AFS group was approximately double that of the IFS group. These results suggest that IFS effectively restricts water mobility. By maintaining a consistent thermal profile, IFS minimizes the external disturbances that drive water desorption. In addition to the changes in water population proportions, the shifts in relaxation time (T2 peak positions) provided deeper insights into the water-substrate interactions. Over the 60-day storage period, the T22 peak (immobilized water) in the AFS group exhibited a noticeable rightward shift toward longer relaxation times (from 11.89 ms to 14.63 ms). A longer relaxation time directly reflects an increased degree of freedom for water molecules, indicating a weakened binding strength between the water and the gluten-starch matrix. This thermodynamic loosening is primarily caused by the mechanical disruption of the protein network by large, recrystallized ice crystals, which destroy the specific hydrogen-bonding sites. Conversely, the IFS group displayed a much smaller rightward shift (reaching only 12.75 ms), confirming that the stable thermal environment effectively maintained the binding affinity between water molecules and the structural macromolecules. (Zhao et al., 2025). Freezable water fraction Changes in water mobility, as revealed by LF-NMR, are closely linked to the amount of water available for phase transition. Fig. 4(c) shows that the content of freezable water (free and loosely bound water) increased with storage duration. This shift is a key indicator of structural damage driven by ice recrystallization. In the AFS group, temperature fluctuations promote the continuous growth of ice crystals. As these crystals enlarge, they mechanically disrupt the gluten network and break the non-covalent forces (such as hydrogen bonds) that hold water to proteins and starch. Consequently, water that was previously bound is released and converted into freezable water (Jiang et al., 2023). Notably, this increase was much less pronounced in the IFS group (reaching 27.82%) compared to the AFS group (38.58%). These results imply that by acting as a thermal buffer, IFS limits the temperature shocks that drive this water conversion. This stability helps preserve the original water-binding capacity of the gluten network, offering a protective effect during long-term storage. Ice crystal morphology and growth behavior Fig. 5(a) illustrates the microstructure of the dough after 60 days. Due to the sample preparation (fixation and dewaxing), the ice crystals appear as white voids within the stained dough matrix. Quantitative analysis (Fig. 5b-d) revealed a progressive enlargement of these voids over time, a clear sign of ice crystal growth. Crucially, the extent of this growth is heavily influenced by the thermal history of the sample.Fig. 5Effects of IFS on ice crystal. (a) Original image of the slices; (b) gray-scale conversion map of the slices; (c) figure of the quantitative analysis of the mean area of ice crystals. AFS: air frozen storage (AFS); IFS: isothermal-buffered frozen storage.Fig. 5 Effects of IFS on ice crystal. (a) Original image of the slices; (b) gray-scale conversion map of the slices; (c) figure of the quantitative analysis of the mean area of ice crystals. AFS: air frozen storage (AFS); IFS: isothermal-buffered frozen storage. In the AFS group, the temperature fluctuations inherent to air-based storage act as a driving force for Ice crystal growth. In this process, temperature oscillations promote the melting of smaller, thermodynamically unstable crystals and their subsequent redeposition onto larger ones (Obadi & Xu, 2024; Zhang et al., 2020). This leads to the formation of large, irregular ice domains that distort the surrounding matrix. By the end of the storage period, the average ice crystal area in the IFS group was 18.49 μm2, which was significantly smaller than that of the AFS group (24.64 μm2, p < 0.05). This difference highlights the efficacy of IFS in suppressing crystal coarsening. By maintaining a stable thermal environment, IFS limits the thermodynamic driving force for recrystallization. This results in smaller, more uniform ice crystals that cause less structural disruption to the dough, a finding consistent with the correlation between thermal stability and crystal size reported by Jiang, Yang, et al. (2025). Gluten network integrity visualized by CLSM To visually assess the structural evolution of the gluten network, Confocal Laser Scanning Microscopy (CLSM) was performed (Fig. 6a, b). Fresh dough displayed a continuous and uniform protein network. However, after 60 days, the network in the AFS group appeared significantly fragmented, with large dark voids representing the spaces previously occupied by ice crystals.Fig. 6Effects of IFS on protein network structure of frozen dough. (a) Figures of original CLSM; (b) transformation diagram for CLSM; (c) figure of average protein length quantitative; (d) figure of protein area share quantitative. AFS: air frozen storage (AFS); IFS: isothermal-buffered frozen storage.Fig. 6 Effects of IFS on protein network structure of frozen dough. (a) Figures of original CLSM; (b) transformation diagram for CLSM; (c) figure of average protein length quantitative; (d) figure of protein area share quantitative. AFS: air frozen storage (AFS); IFS: isothermal-buffered frozen storage. Based on these structural observations, we speculate that the extensive protein fragmentation in the AFS group might be driven by a combination of direct physical damage and subsequent molecular alterations. First, direct physical disruption plays a primary role. As unavoidable temperature fluctuations promote ice recrystallization, the growing, irregular ice crystals can act as physical wedges that mechanically sever the continuous protein strands (Liu et al., 2025). Second, this mechanical stress triggers biochemical depolymerization. The sharp ice crystals also puncture yeast cell membranes, leading to the release of endogenous reducing agents (like glutathione). As reported by previous studies (Cui et al., 2023; Silvas-García et al., 2016), these agents can chemically cleave disulfide bonds and degrade the polymeric glutenin network. Finally, the repeated freeze-dehydration induced by thermal fluctuations is likely to bring protein chains into unnatural proximity, potentially promoting hydrophobic over-aggregation into non-functional clumps. Quantitative analysis (Fig. 6c, d) confirms that the IFS group retained a much stronger network. After 60 days, the protein area fraction in the IFS group was 7.96% higher than that of the AFS group, and the average protein length was 86.86 μm longer. By providing a more stable thermal environment, IFS appears to alleviate excessive water migration and thermal stress on yeast cells. We therefore hypothesize that this buffering effect helps mitigate the direct physical severing by ice crystals, and further delays both biochemical depolymerization and physical over-aggregation, which partially explains the more intact and continuous network visualized in the CLSM images. Microstructural alterations of the starch-gluten matrix During frozen storage, the stability of the starch-gluten matrix is heavily influenced by the thermal environment. In the AFS group, inherent temperature fluctuations create a harsh environment for the dough microstructure, which appears to drive starch-gluten detachment through both physical and chemical pathways. Physically, these fluctuations promote ice recrystallization, leading to the growth of large crystals that can act as physical wedges, mechanically forcing the starch granules to separate from the gluten network. Additionally, the repeated temperature changes cause cycles of volume expansion and contraction. Since starch and protein have different thermal properties, this mismatch generates mechanical stress at their interface, further promoting structural detachment (Zhang et al., 2007). Chemically, the continuous thermal agitation and subsequent water migration can disrupt the interfacial hydration layer. This localized dehydration is likely to weaken the non-covalent interactions, particularly hydrogen bonds, that facilitate the natural adhesion between the starch granule surface and the protein matrix(Liu et al., 2024). Microstructural observations (Fig. 7) clearly reflect this impact. In fresh dough, starch granules were tightly wrapped and uniformly embedded within the gluten network. However, after 60 days of AFS, a significant number of starch granules became exposed and detached from the protein matrix, with some appearing deformed. In contrast, the IFS group maintained a much more intact structure, with fewer exposed granules and better adhesion between starch and gluten. This suggests that by buffering temperature changes, IFS helps mitigate not only the physical mechanical forces that pry the starch and gluten apart, but also the chemical destabilization of their interfacial bonds. This thermal stability is crucial for preserving the integrity of the starch-protein composite (Jiang et al., 2025; Liu et al., 2024).Fig. 7Effects of IFS on microstructure of frozen dough. AFS: air frozen storage; IFS: isothermal-buffered frozen storage. A7, A15, A30, and A60 represent AFS samples stored for 7, 15, 30, and 60 days, respectively, while I7, I15, I30, and I60 represent IFS samples stored for the corresponding durations.Fig. 7 Effects of IFS on microstructure of frozen dough. AFS: air frozen storage; IFS: isothermal-buffered frozen storage. A7, A15, A30, and A60 represent AFS samples stored for 7, 15, 30, and 60 days, respectively, while I7, I15, I30, and I60 represent IFS samples stored for the corresponding durations. Multivariate statistical analysis of quality parameters To systematically explore the relationships among physicochemical properties, water status, and microstructural characteristics, Pearson correlation analysis and Principal Component Analysis (PCA) were conducted. The Pearson correlation heatmap (Fig. 8a) visually summarizes the associations between these parameters. A strong negative correlation was observed between the average ice crystal area and key quality indicators, such as specific volume, fermentation capacity, and bound water content. Conversely, ice crystal area showed a strong positive correlation with proofing time and freezable water content. Statistically, this highlights that ice crystal growth is the central driver of quality loss. Since, as discussed in previous sections, ice crystal growth is significantly accelerated by the temperature fluctuations in AFS, these correlations statistically reinforce the critical importance of thermal stability in preserving dough quality (Obadi & Xu, 2024).Fig. 8Pearson correlation analysis (a) and Principal Component Analysis (b) of quality attributes and structural parameters in frozen dough during storage. AFS: air frozen storage; IFS: isothermal-buffered frozen storage. A7, A15, A30, and A60 represent AFS samples stored for 7, 15, 30, and 60 days, respectively, while I7, I15, I30, and I60 represent IFS samples stored for the corresponding durations.Fig. 8 Pearson correlation analysis (a) and Principal Component Analysis (b) of quality attributes and structural parameters in frozen dough during storage. AFS: air frozen storage; IFS: isothermal-buffered frozen storage. A7, A15, A30, and A60 represent AFS samples stored for 7, 15, 30, and 60 days, respectively, while I7, I15, I30, and I60 represent IFS samples stored for the corresponding durations. Principal Component Analysis (PCA) further differentiated the effects of the two storage treatments (Fig. 8b). The first principal component (PC1) served as the primary dimension for distinguishing sample quality. Variables associated with fresh dough (e.g., specific volume, moisture content) loaded negatively on PC1, while indicators of deterioration (e.g., ice crystal area, freezable water) loaded positively. In the score plot, both groups shifted along the PC1 axis over time, indicating natural aging. However, a distinct separation was evident by the end of the 60 days, the IFS sample (I60) was positioned significantly closer to the fresh control than the AFS sample (A60). This spatial separation suggests that the thermal stability provided by IFS effectively slowed down the deterioration trajectory. In contrast, the AFS group drifted further toward the deterioration dimension, driven by the cumulative stress of temperature fluctuations and recrystallization (Diao et al., 2021). To explore potential feature vectors for measuring the degree of freezing deterioration, the factor loadings along the first principal component (PC1) were further examined based on the biplot. The analysis indicated that deterioration indicators, such as mean ice crystal area, freezable water content, and mean stomatal area, exhibited substantial positive contributions to PC1. Conversely, quality-preserving indicators, including specific volume, bound water, and mean protein length, showed prominent negative correlations. These specific parameters appear to align with the trajectory from fresh to deteriorated states, suggesting they could serve as relevant markers for evaluating frozen dough deterioration in this study. Underlying mechanisms of quality deterioration and preservation Based on the statistical correlations and microstructural evidence presented above, we propose a potential mechanism to explain the differential effects of AFS and IFS (illustrated in Fig. 9). The divergence in product quality appears to stem primarily from the distinct thermal stability of the two storage methods. In the AFS group, the low thermal conductivity of air makes it difficult to maintain a constant temperature. The resulting temperature fluctuations act as a trigger for a synergistic deterioration cascade involving water migration, the gluten network, and yeast activity. First, these fluctuations cause small ice crystals to melt and refreeze onto larger ones. These growing, irregular crystals can act as physical wedges, mechanically disrupting the gluten network and potentially inducing cryo-injury to yeast cells. As discussed previously, this damage to yeast triggers the release of endogenous reducing agents (e.g., glutathione), which chemically depolymerize the gluten proteins. The combination of physical rupture and chemical depolymerization significantly compromises the water-binding sites on proteins and starch. Consequently, previously bound water tends to be released and converted into free water. This newly freed water is then available to join the ice phase during the next temperature drop, further fueling crystal growth and completing this synergistic feedback loop.Fig. 9Schematic illustration of the proposed mechanism underlying the differential effects of AFS and IFS on frozen dough quality. AFS: air frozen storage; IFS: isothermal-buffered frozen storage.Fig. 9 Schematic illustration of the proposed mechanism underlying the differential effects of AFS and IFS on frozen dough quality. AFS: air frozen storage; IFS: isothermal-buffered frozen storage. In comparison, IFS appears to interrupt this loop. The liquid medium, with its high heat capacity, acts as a steady-state thermal buffer. It helps shield the dough from external temperature changes, thereby inhibiting the recrystallization process. By maintaining a stable thermal environment, IFS keeps ice crystals relatively small and uniform. This minimizes the initial mechanical stress on the gluten network and yeast cells, which in turn assists in keeping water tightly bound within the matrix. By mitigating the conversion of bound water to free water, IFS helps sustain the structural integrity and fermentation capacity of the dough, ultimately contributing to the improved bread volume and texture observed in this study. Conclusion This study demonstrates that isothermal-buffered frozen storage (IFS) effectively mitigates the long-term deterioration of frozen dough. By dampening temperature fluctuations, IFS limits ice recrystallization and preserves the gluten-starch network, thereby interrupting the synergistic deterioration cascade among water, protein, and yeast cells. These findings highlight that maintaining long-term thermal stability is as critical as the initial freezing rate. However, the current liquid-immersion IFS system remains an initial experimental exploration, its reliance on strict vacuum packaging poses logistical limitations for large-scale industrial applications. Future efforts should focus on translating this buffering principle into scalable formats, such as integrating phase-change materials into smart packaging, to provide practical solutions for real-world cold chain logistics. CRediT authorship contribution statement Shuang Li: Writing – original draft, Methodology, Investigation, Formal analysis, Data curation. Zixuan Wu: Methodology, Data curation. Xianghong Li: Methodology, Investigation. Yiqun Huang: Visualization, Resources. Yongle Liu: Writing – review & editing, Conceptualization. Faxiang Wang: Writing – review & editing, Validation, Supervision, Project administration, Methodology, Funding acquisition, 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.