Mechanistic Insights into the Cooperative Removal of NH3 and H2S by Persimmon Polyphenols with Natural Deep Eutectic Solvent Systems

TL;DR

This study explores how persimmon polyphenols, when combined with natural deep eutectic solvents, can effectively remove ammonia and hydrogen sulfide odors through specific chemical interactions.

Contribution

The novel use of natural deep eutectic solvents to enhance persimmon polyphenols' odor-removal capabilities through cooperative microenvironment regulation.

Findings

Chloride-citric acid NADES was optimal for persimmon polyphenol extraction and ammonia removal.

Betaine-urea NADES was more effective for hydrogen sulfide removal due to acid-base and hydrophobic interactions.

NADESs enhance deodorization by regulating microenvironments rather than causing irreversible chemical changes.

Abstract

Persimmon polyphenols (PP) are natural polyphenols with high reactivity and strong deodorization potential; however, their practical application in odor control is limited by their poor solubility. In this study, natural deep eutectic solvents (NADESs) were employed for the green extraction of PP, and the capabilities of extracts on the removal of ammonia (NH3) and hydrogen sulfide (H2S) were investigated. In addition, the underlying mechanisms were explored by integrating spectroscopic analysis, molecular dynamics simulations, and quantum chemical calculations. The results showed that chloride-citric acid (CC-CA) was the optimal system in both PP extraction and sustained NH3 removal, while the betaine-urea (B-U) system was more effective for H2S removal. NH3 removal was governed by acid-base neutralization, with the resulting ammonium species being further stabilized within the PP-regulated NADES hydrogen-bond network. In contrast, H2S interacted with the solvent network not only through acid-base neutralization but also via Van der Waals forces and hydrophobic contacts. Our data supported that NADESs enhanced the deodorization performance of PP through cooperative microenvironment regulation rather than irreversible chemical conversion. This work highlighted that NADESs could not only function as highly efficient extraction media for polyphenols, but also active platforms for enhancing selective gas-capture capability for polyphenols. Furthermore, it provided a new strategy for the rational design of green, persimmon-derived deodorants.

Figures

Full text

1. Introduction

Persimmon (Diospyros kaki L.), a species within the genus Diospyros of the Ebenaceae family, is one of the most economically significant fruit trees in East Asia [1]. The fruits are particularly rich in polyphenolic compounds, collectively referred to as persimmon polyphenols (PP), which are traditionally classified into soluble and insoluble fractions [2] based on their physicochemical behaviors. Soluble PP, composed of monomeric and oligomeric phenolics such as gallic acid and catechins [3], confer the characteristic astringency of unripe fruits. During maturation, soluble PP undergo enzymatic oxidation and condensation, polymerize into high-molecular-weight structures, and become integrated into the cell wall matrix [4], eventually forming insoluble PP. In addition to mature fruits, thinned green persimmons [5], which are usually discarded or composted, also contain high levels of both soluble and insoluble tannins, making them valuable sources of polyphenols and providing a straightforward route for valorizing this agricultural byproduct.

Besides the well-known diverse biological activities, including antioxidant effects, cholesterol-lowering effects, and antibacterial properties [6], recent studies have also highlighted the potential of polyphenols in the absorption and removal of toxic gases. Polyphenol-based materials can attenuate NH_3_ emissions through hydrogen bonding, protonation, and adsorption [7]. In sulfur-containing systems, polyphenolic extracts also contribute to H_2_S removal via coordinated adsorption and redox-assisted desulfurization pathways [8]. The efficacy of persimmon tannin against volatile sulfur compounds (VSCs), such as methyl mercaptan and hydrogen sulfide, has been well established in the context of oral malodor [9]. It is well known that polyphenols from green tea or persimmon are effective in removing body odor, especially for elderly individuals [10]. The multivalent binding capacity of polyphenols is derived from their dense phenolic hydroxyl groups and conjugated aromatic domains. This structural foundation facilitates multiple gas-capture mechanisms, as established in previous studies on polyphenol–gas interactions [11,12]. Collectively, these characteristics enable polyphenols to neutralize malodorous gases such as NH_3_ and H_2_S through a combination of complementary non-covalent interactions and redox-related pathways.

Compared to conventional deodorants, such as strong oxidants and physical adsorbents, which often suffer from drawbacks including secondary pollution, limited adsorption and regeneration capacity, and low selectivity toward specific odorants, natural polyphenols are novel, green and environmentally friendly absorbents for toxic gas capture. However, the practical application of polyphenols in toxic gas absorption is often limited by their relatively poor water solubility [13]. Limited solubility can hinder the exposure of reactive phenolic sites, thereby reducing deodorization performance and ultimately restricting the sustainable and efficient use of polyphenols.

Natural deep eutectic solvents (NADESs), formed through extensive hydrogen bonding between hydrogen-bond donors (HBDs) and acceptors (HBAs), offer a green solvent platform with highly tunable polarity, acidity/basicity, and network structures [14]. Beyond their well-established applications in extracting plant constituents like proanthocyanidins, pigments, and polysaccharides, emerging evidence indicates that certain NADESs possess intrinsic chemical reactivity, which enables them to act as promising absorbents for capturing toxic gases, including NH_3_, H_2_S, CO_2_, and SO_2_ [15,16,17]. Such gas-capture ability underscores that NADESs can function as both efficient extraction media and active platforms for environmental and chemical applications [18,19]. This prompted us to propose that, by designing appropriate NADES systems, PP could not only be extracted efficiently in a green manner, but its toxic gas-capture capability could also be cooperatively enhanced. However, current studies on the selective removal of NH_3_ and H_2_S by different NADES-PP combinations remain scarce.

Therefore, in the present study, based on the polyphenol profile analysis of thinned green persimmons, appropriate NADES systems were screened for both PP extraction and the selective removal of NH_3_ and H_2_S. Furthermore, the synergistic effects arising in NADES-PP systems were investigated. Molecular dynamics (MD) simulations of representative systems were also conducted to reveal the molecular-scale interactions with NH_3_ and H_2_S. This integrated approach provides mechanistic insights into gas capture and aims to establish a theoretical basis for the development of efficient, persimmon-derived deodorants, ultimately supporting the expanded industrial utilization of persimmon-based materials.

2. Materials and Methods

2.1. Chemicals and Reagents

Green persimmons (Diospyros kaki L.) at maturity stages 4–5 were harvested from the Flower Base of Huazhong Agricultural University (Wuhan, China). After stem removal and washing, the fruits were thinly sliced, freeze-dried (model LGJ-10, Beijing Songyuan Huaxing Technology Development Co., Ltd., Beijing, China) for 60 h, and ground into a fine powder using a pulverizer (model LG-04, Ruian Baixin Pharmaceutical Machinery Co., Ltd., Ruian, China).

The following components were used for the preparation of NADESs: betaine (B), choline chloride (CC), L-proline (P), citric acid (CA), lactic acid (LA), malic acid (MA), glycerol (GL), urea (U), and 1,2-butanediol (BD). Other chemical reagents, including anhydrous sodium carbonate, formaldehyde solution, ammonia solution (GR, 25.0–28.0%), hydrochloric acid, methanol (analytical grade), and ethanol (analytical grade), were procured from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). Folin–Ciocalteu reagent and potassium bromide (spectroscopic grade) were obtained from Shanghai Yuanye Biotech Co., Ltd. (Shanghai, China).

All reference standards, gallic acid (GA), catechin (C), epicatechin (EC), epigallocatechin gallate (EGCG), and procyanidin B, along with trifluoroacetic acid (HPLC grade), methanol (HPLC grade), acetonitrile (HPLC grade), formic acid (MS grade), sodium sulfide, and citric acid, were obtained from McLean Biochemical Technology Co., Ltd. (Shanghai, China). Ultrapure water, prepared using a laboratory water purification system, was used throughout all experiments.

2.2. Preparation of NADESs

NADESs were prepared following a previously reported method with slight modifications [20]. Briefly, the hydrogen-bond acceptor (HBA), hydrogen-bond donor (HBD), and water were accurately weighed according to the molar ratios specified in Table 1 and combined in a glass vessel. The mixture was magnetically stirred at 80 °C until a clear, homogeneous liquid formed. The resulting NADESs were stored at 25 °C for subsequent use.

2.3. Extraction of PP

The extraction of PP was performed with slight modifications based on our previous studies [21]. Freeze-dried persimmon powder was first mixed with 10% (v/v) ethanol at a solid-to-liquid ratio of 1:10 (g/mL). The suspension was magnetically stirred at 400 rpm for 1 h at room temperature under light-protected conditions and then allowed to stand for an additional hour in the dark. The mixture was filtered through a 100-mesh sieve, and the retained solids were collected. This solid residue was freeze-dried for 60 h and ground to obtain desugared persimmon powder. Based on the moisture content (80%) and sugar content (11.2%) of fresh persimmon, the final yield of desugared powder was calculated to be 8.8%.

The desugared powder was then mixed with the extraction solvent at a ratio of 1:35 (g/g), and extracted in an 80 °C water bath for 40 min. This extraction process was repeated three times. The combined filtrates were centrifuged at 10,000 rpm for 10 min, and the resulting supernatant was collected for further analysis.

2.4. UPLC-ESI-QTOF-MS Analysis

Qualitative analysis of chemical components was performed using a Waters ACQUITY UPLC I-Class PLUS system coupled with a Xevo G2 XS QTOF mass spectrometer equipped with an electrospray ionization (ESI) source. Chromatographic separation was achieved on a Waters ACQUITY UPLC BEH C_18_ column (2.1 × 100 mm, 1.7 μm) maintained at 35 °C. The mobile phase consisted of 0.13% formic acid in water (A) and 0.1% formic acid in acetonitrile (B), with a gradient elution program at a flow rate of 0.17 mL/min. Mass detection was conducted in negative ESI mode over an m/z range of 10–1000. Detailed chromatographic and mass spectrometric conditions are provided in Table A1.

2.5. Total Phenolic Content (TPC)

The total phenolic content (TPC) was determined using the Folin–Ciocalteu method [22]. Briefly, 0.2 mL of the sample was mixed with 0.5 mL of Folin–Ciocalteu reagent and allowed to react in the dark for 3 min. Then, 3 mL of sodium carbonate solution (7.5% w/w) was added. The mixture was thoroughly vortexed and then incubated in the dark at room temperature for 1 h. The absorbance was measured at 760 nm using a UV–vis spectrophotometer. A calibration curve was constructed using GA, and results were expressed as mg of GA equivalents per gram of fresh weight (GAE/g FW).

2.6. Deodorization Capacity Assay

Preliminary kinetic experiments were performed to determine the time required for gas–liquid equilibrium. Representative NADES-PP systems (CC-CA-PP for NH_3_, B-U-PP for H_2_S) were sampled every 2 min from 2 to 30 min after sealing. Based on the manufacturer’s instructions for the detector tubes and experimental observations, headspace concentrations stabilized within 30 min for both gases (coefficient of variation <5% between 25 and 30 min). Thus, a 30 min equilibration time was selected for all experiments.

Deodorization performance of the samples was evaluated by measuring the removal capacity of NH_3_ and H_2_S. Gaseous NH_3_ was generated from a 0.4% (v/v) ammonia solution, while H_2_S was released from a 0.1% (w/w) sodium sulfide solution acidified to pH 4–5 with citric acid. In a 250 mL sealed conical flask, 1 mL of the odor source solution was added and allowed to equilibrate for 1 min. Subsequently, 1 mL of the deodorant solution was added. A blank control was prepared by replacing the deodorant solution with an equal volume of distilled water. All samples were maintained at 25 °C for 30 min.

Gas concentrations were measured using gas detector tubes (Kitagawa Precision Detector Tubes, Komyo Kitagawa, Tokyo, Japan) in a sealed batch test. For NH_3_ measurement: Kitagawa Tube No. 105SC (detection range: 5–260 ppm, detection limit: 0.05 ppm, relative standard deviation: ±10% at low concentrations and ±5% at mid-to-high concentrations according to manufacturer specifications). For H_2_S measurement: Kitagawa Tube No. 120SB (detection range: 3–150 ppm, detection limit: 0.3 ppm, relative standard deviation: ±10% at low concentrations and ±5% at mid-to-high concentrations according to manufacturer specifications).

All detector tubes were from the same manufacturing lot and used within their expiration dates. Prior to each experimental series, the sampling pump was verified for airtightness and according to the specifications of manufacturer. A fresh detector tube was used for each individual measurement, and each sample was analyzed in triplicate using three separate tubes. Reproducibility was assessed by performing five consecutive measurements on identical samples under the same conditions. The coefficient of variation (CV) was <5% for both NH_3_ and H_2_S measurements, indicating good reproducibility. The selectivity of the detector tubes was evaluated by measuring the response of each tube to the non-target gas under the same experimental conditions described above. Specifically, the NH_3_ detector tube was exposed to H_2_S gas in the absence of NH_3_, and the H_2_S detector tube was similarly exposed to NH_3_ gas. Neither detector tube showed measurable response to the non-target gas (both readings < detection limit), confirming that cross-interference is negligible under our experimental conditions.

All deodorization assays were conducted in 250 mL conical flasks with a calibrated total volume of 285 ± 3 mL (determined by water displacement). Each flask contained a fixed liquid volume of 2 mL (1 mL odor source + 1 mL deodorant solution), resulting in a standardized headspace volume of approximately 283 mL. Flasks were immediately sealed with Parafilm and screw caps to ensure airtight conditions.

All experiments were performed in triplicate. The removal rate was calculated using the following formula:

$$[eqn]$$

where c_0_ is the initial gas concentration in the blank control, and c_t_ is the residual gas concentration after the reaction.

2.7. Optimization of NADES

Following the initial screening, two NADES systems were selected for further study: CC-CA for NH_3_ removal and B-U for H_2_S removal.

2.7.1. Extraction Performance Optimization

Based on the initial screening results, the molar ratio of HBA:HBD (CC:CA), the molar ratio of H_2_O:HBD (H_2_O:CA), and the solvent-to-solid ratio were identified as critical factors influencing the PP extraction yield and were selected for optimization. Single-factor experiments were first conducted to assess the individual effects of these variables on total phenolic content (TPC). Subsequently, a three-factor, three-level Box-Behnken design (BBD) was employed for response surface methodology (RSM) optimization. The model was used to investigate the interaction effects among the variables. Experimental validation under the predicted optimal conditions was conducted to verify the accuracy of the model.

2.7.2. Deodorization Performance Optimization

Given that the B-U system demonstrated superior synergy with PP for H_2_S removal but exhibited suboptimal extraction capacity, optimization was focused exclusively on enhancing deodorization performance.

The effects of two key parameters, the HBA:HBD (B:U) molar ratio and the H_2_O:HBD (H_2_O:U) molar ratio, on the H_2_S removal rate were investigated. These ratios were varied systematically while keeping other deodorization assay conditions constant. The system exhibiting the highest H_2_S removal rate was identified as the optimal B-U formulation.

2.8. pH Monitoring During Gas Capture

The pH variation of NADES-PP systems during gas capture was monitored to assess changes in system acidity. Measurements were performed using a digital pH meter equipped with a microelectrode (FiveEasy Plus FE28, Mettler Toledo, Greifensee, Switzerland). Prior to analysis, the pH meter was calibrated using a three-point calibration method with standard buffer solutions at pH 4.01, 7.00, and 9.21.

For each measurement, a defined volume of the sample was placed in a centrifuge tube and mixed with an equal volume of the odorous source solution. The pH was monitored over a 30-min period at room temperature by immersing the electrode directly into the liquid phase, with data collected at 2-min intervals.

2.9. FT-IR Analysis

Fourier-transform infrared (FT-IR) spectra were obtained to investigate the interaction mechanisms among NADESs, persimmon polyphenols (PP), and odorous gases. Spectra of representative NADES systems, their corresponding PP-containing mixtures, and samples after exposure to NH_3_ or H_2_S were collected. Measurements were performed using a Nicolet iS50 FT-IR spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) with the KBr pellet method over the range of 4000–400 cm^−1^ at a resolution of 4 cm^−1^. Each spectrum was obtained by accumulating 32 scans. After background correction, shifts and intensity changes of characteristic absorption bands were analyzed.

2.10. Molecular Dynamics (MD) Simulations

Molecular structures of EGCG, choline cation, chloride anion, citric acid, betaine, and urea were optimized at the B3LYP/6–31G* level using GaussView 05 and Gaussian. Topology files were generated with Sobtop 1.0 [23], and partial charges were assigned using the RESP method via Multiwfn [24,25].

MD simulations were performed using GROMACS 2020.6. Four systems (12 nm cubic boxes) were constructed with Packmol [26]: (i) CC-CA-EGCG/NH_3_ (2262 Ch, 2262 Cl^-^, 1009 CA, 34,115 H_2_O, 12 EGCG, 30 NH_3_); (ii) CC-CA-EGCG reference (4524 Ch, 4524 Cl^−^, 2018 CA, 12,778 H_2_O, 23 EGCG); (iii) B-U-EGCG/H_2_S (1285 B, 2570 U, 44,274 H_2_O, 12 EGCG, 7 H_2_S); (iv) B-U-EGCG reference (2570 B, 5140 U, 30,843 H_2_O, 23 EGCG).

The Amber ff99SB-ildn and GAFF force fields were applied, with TIP3P water. All systems were energy-minimized using the steepest descent algorithm under periodic boundary conditions. Equilibration involved 1 ns NVT (300 K, Berendsen thermostat) followed by 1 ns NPT (300 K, 1.0 bar, Berendsen barostat). Production runs were conducted for 120 ns under NPT with a 2 fs time step. Hydrogen bonds were constrained using LINCS. Long-range electrostatics were treated with the PME method (1.2 nm cutoff).

Trajectory analysis and visualization were performed using VMD [27]. Average non-covalent interactions were analyzed with the amIGM approach in Multiwfn based on the RDG method [28,29].

2.11. Statistical Analysis

All experiments were conducted with at least three independent replicates, and data are expressed as mean ± standard deviation (SD). Prior to analysis, normality was assessed using the Shapiro-Wilk test, and homogeneity of variances was verified using Levene’s test. All data met the assumptions of normality (p > 0.05) and homogeneity of variance (p > 0.05). Statistical significance was determined by one-way analysis of variance (ANOVA), followed by Duncan’s multiple range test for post-hoc comparisons, with a significance level defined at p < 0.05. Effect sizes were reported as partial eta-squared (η^2^) for ANOVA results, and 95% confidence intervals were calculated for mean differences where applicable. Data processing and ANOVA were performed using SPSS software (version 23.0; IBM, USA). The Box-Behnken design (BBD) generation and corresponding response surface methodology (RSM) statistical analyses were completed using Design-Expert software (version 13; State-East Inc., Palmer, MA, USA).

3. Results and Discussion

3.1. Identification of Phenolic Compounds in Unripe Persimmons

To determine the specific composition of soluble PP, their ethanol extract underwent UPLC-Acquity-TOF/MS analysis. As shown in Figure 1A and Table 2, a total of 13 polyphenols were identified. Structural identification of these compounds was based on their precursor ion and fragment ion peak information, combined with literature reports and MS database comparisons, and some compounds were confirmed by matching retention times and fragmentation patterns with reference standards.

As summarized in Table 2, peak 1 exhibited a deprotonated molecular ion at m/z 169.0135 ([M-H]^−^) with a retention time consistent with that of the authentic standard. The MS/MS spectrum was dominated by a fragment ion at m/z 125.0243, corresponding to the neutral loss of CO_2_ (44 Da). This characteristic decarboxylation is diagnostic of gallic acid and its derivatives, confirming the identity of this compound as gallic acid.

Peak 2 displayed an [M-H]^−^ ion at m/z 305.0659 and generated two fragment ions at m/z 241.8971 and 125.0233, separately. The fragment at m/z 241.8971 likely resulted from the elimination of an o-cresol moiety, while the signal at m/z 125.0233 originated from subsequent A-ring cleavage, a fragmentation pathway typical of flavan-3-ols. The relatively low response intensity and limited number of fragment ions suggested that the A- and C-ring structures remained comparatively stable during collision-induced dissociation. Combined with its retention behavior and reported fragmentation patterns, peak 2 was identified as epigallocatechin.

Peak 3 and peak 4 showed identical precursor ions at m/z 289.0725 ([M-H]^−^), indicating that they were structural isomers. Their MS/MS spectra yielded common fragment ions at m/z 245.0825, 125.0231, and 109.0263, produced through retro-Diels-Alder (RDA) cleavage of the flavan-3-ol skeletons. Comparison with authentic standards and the literature data allowed their differentiation; peak 3 and peak 4 were assigned to catechin and epicatechin, separately.

Peak 5 displayed a parent ion at m/z 183.0294 ([M-H]^−^), accompanied by a prominent fragment at m/z 124.0166, a diagnostic ion commonly observed in methylated gallic acid derivatives [30]. Its fragmentation pattern and retention profile supported the tentative assignment as methyl gallic acid.

For peak 6, an [M-H]^−^ ion was observed at m/z 457.0770, matching both the retention time and MS/MS profile of the epigallocatechin gallate (EGCG) standard. Fragment ions at m/z 305.0531 and 287.0434 reflected successive losses of galloyl groups, characteristic of gallated catechins. This compound was therefore unequivocally identified as EGCG.

An additional precursor ion at m/z 577.1346 appeared as several partially overlapping peaks, suggesting oligomeric flavanol species. The corresponding MS/MS spectrum exhibited a major fragment at m/z 287.1385, indicative of an epicatechin monomer generated by interflavan bond cleavage. This fragmentation pattern is consistent with procyanidin B, a typical procyanidin dimer.

Peak 8 and peak 9 displayed deprotonated molecular ions at m/z 463.0818 and 463.0873, respectively, indicating that they are isomeric quercetin glycosides. Their MS/MS spectra were both characterized by the neutral loss of a hexose unit (162 Da), yielding the diagnostic quercetin aglycone ion at m/z 301. Differences in retention time implied distinct sugar linkages or configurations. Based on these features and literature comparisons, peak 8 and peak 9 were identified as hyperoside and isoquercitrin, respectively [1].

Peak 10 gave a precursor ion at m/z 447.0925, with fragmentation producing ions at m/z 300.9016 and 284.0293. The signal near m/z 301 corresponded to the quercetin aglycone after deoxyhexose loss, while the m/z 284 ion likely arose from further CO elimination or ring contraction. This fragmentation behavior aligned with quercetin glycosides containing deoxyhexose units; thus, the compound was assigned as a quercetin glycoside [31].

Peak 11 displayed an [M-H]^−^ ion at m/z 433.0766, consistent with a quercetin–pentose conjugate. The major fragment ion at m/z 301 reflected a neutral loss of 132 Da (arabinose), matching reported spectra for quercetin-3-O-arabinoside [32].

Peak 12 produced an [M-H]^−^ ion at m/z 615.1003, giving sequential fragments at m/z 463.0881 and 301.0331, corresponding to the stepwise loss of a galloyl group (152 Da) followed by a hexose moiety (162 Da). Such two-stage fragmentation is typical of galloylated quercetin glycosides, leading to its tentative identification as quercetin-3-O-(2″-O-galloyl)-β-D-glucoside [33].

Peak 13 presented a deprotonated ion at m/z 301.0362, with a characteristic fragment at m/z 273.8271 produced via RDA cleavage of the C-ring under negative mode. This fragmentation pattern is diagnostic for quercetin, confirming the identity of the compound as the quercetin aglycone [34].

The HPLC analysis results for insoluble PP are shown in Figure 1B. By comparison with authentic standards, the sample was found to contain gallic acid, epigallocatechin gallate (EGCG), and condensed tannins. The distinct peak at approximately 5 min corresponded to gallic acid, while the sharp peak around 10 min matched the retention time of the EGCG standard. In contrast, a broad and continuous peak cluster observed between 10 and 25 min, characterized by irregular peak shapes and extended retention times, was attributed to condensed tannins. Such high-molecular-weight polyphenols commonly eluted as tailing or overlapping peaks on reversed-phase C18 columns due to their pronounced structural heterogeneity and strong hydrophobicity [35]. Consistent with these chromatographic characteristics, the structural composition of persimmon tannins has been systematically elucidated in our previous work. Thiolytic degradation analysis revealed that condensed tannins from Chinese persimmons were mainly composed of flavan-3-ol subunits, including epicatechin (EC), epigallocatechin (EGC), epicatechin-3-O-gallate (ECG), and EGCG as extension units, with myricetin, catechin (C), and EGCG acting as terminal units [36,37].

Collectively, these results indicated that EGCG represented both a structurally important component and representative sub-unit within both soluble and insoluble fractions.

3.2. Screening NADES from Both Extracting Capacity and Deodorization Performance

The capacities of NADES in extracting PP from unripe persimmons were firstly evaluated based on total phenolic content (TPC), with a conventional 1% HCl-methanol system serving as a reference. As shown in Figure 2, the PP extraction capacities of NADES were significantly (p < 0.05) affected by the HBA and HBD compositions. Among all the combinations, the CC-CA system exhibited the highest extraction (9.57 ± 0.01 mg GAE/g FW) capacity, comparable to the conventional HCl-methanol reflux method (9.68 ± 0.06 mg GAE/g FW), indicating that CC-CA demonstrated a strong potential to replace acidified organic solvents for PP extraction, while the CC-U system showed the poorest extraction performance (only 0.14 ± 0.01 mg GAE/g FW). Notably, when the hydrogen bond acceptor was identical, NADESs composed of organic acids such as CA, LA, and MA exhibited superior extraction efficiencies (7.0–9.6 mg GAE/g FW) compared with those containing neutral or alkaline hydrogen-bond donors, including glycerol or urea (0.1–7.0 mg GAE/g FW). The superior performance of acid-based NADESs could be attributed to multiple factors [38]. Firstly, organic acids created an acidic micro-environment and thus facilitated the PP extraction by weakening hydrogen bonding interactions between phenolic hydroxyl groups of PP and hydroxyl groups provided by cell wall polysaccharides. In addition, cooperative hydrogen-bond networks formed between PP and choline chloride or organic acids further enhanced the solubilization of polyphenolic structures. By contrast, systems such as CC-U exhibited limited extraction efficiency due to insufficient acidity and strong internal hydrogen bonding, which restricted their interactions with PP molecules. Overall, these results demonstrated that NADES composition, particularly acidity and hydrogen-bonding capability, played a decisive role in PP extraction efficiency, with CC-CA emerging as the most promising green solvent for unripe PP extraction.

The deodorization performance of NADESs toward NH_3_ and H_2_S, both before and after compounding with PP, was systematically evaluated, as summarized in Table 3. To evaluate the deodorization performance of different NADES systems across a measurable and comparable range, preliminary experiments were conducted to identify concentration conditions under which each system exhibited quantifiable removal without complete saturation or negligible response. Based on these pretests, three dilution ratios (1×, 100×, and 1000×) were selected to ensure that removal efficiencies fell within the dynamic measurement range of the detector tubes. Importantly, all experiments were conducted under identical headspace volume, gas generation conditions, and equilibration time. Removal efficiency was calculated as a percentage relative to the corresponding blank control, thereby allowing valid comparison of deodorization performance across different NADES systems despite variation in dilution levels.

To elucidate the intrinsic contribution of PP, water was included as a control solvent. As shown, when PP was dispersed in water, the NH_3_ removal rate was limited (29.17 ± 3.62%). In contrast, NADES-based systems demonstrated substantially enhanced deodorization performance, highlighting the critical role of the solvent matrix in activating and amplifying the functionality of PP. For NH_3_ removal, NADESs formulated with carboxylic acid hydrogen-bond donors, including CA, LA, and MA, exhibited strong intrinsic removal capacities (70–100%, 100-fold dilution). This behavior could be primarily attributed to efficient acid-base neutralization and stabilization of NH_3_ as ammonium species within the NADES network [39]. While systems containing neutral or weakly basic hydrogen-bond donors, including CC-GL, CC-U, and B-U, displayed markedly lower NH_3_ removal capacities (−58–70%, undiluted), with some systems even showing negative removal values. This behavior might be attributed to the limited stabilization capacity of neutral or weakly basic NADESs, which lacked sufficient acidic sites to effectively retain NH_3_. Under these conditions, minor fluctuations in gas–liquid equilibrium during measurement could result in slightly higher detected concentrations relative to the blank control, yielding apparent negative removal rates. These values indicated negligible capture performance. Upon PP addition, a pronounced synergistic enhancement in NH_3_ removal was observed across most NADESs. Unlike the aqueous system, where PP exhibited limited removal capacity (29.17 ± 3.62%, undiluted), PP incorporation into NADESs led to substantially higher NH_3_ removal under identical or even higher dilution conditions. Consequently, several systems, including CC-CA, CC-BD, B-MA, and P-LA, achieved nearly complete NH_3_ removal (100%, 100-fold dilution). This observation suggested that efficient NH_3_ capture arose from the combined contributions of NADESs and PP, in which NADESs providing the primary capture capacity while PP incorporation further reinforced and stabilized the removal capacity.

The removal behavior of H_2_S differed markedly from that of NH_3_. In water, PP exhibited an extremely low H_2_S removal rate (5.13% ± 4.44% undiluted), indicating limited interaction between H_2_S and phenolic hydroxyl groups under these conditions. Without PP addition, urea-based systems such as CC-U and B-U showed substantially high H_2_S removal capacities (CC-U: 71.43% ± 2.38%, undiluted; B-U: 50.79% ± 1.80%, undiluted). In contrast, all other systems exhibited inherently low H_2_S removal rates (−14.44–16.67%, undiluted), with the negative values mainly attributable to measurement variability near the detection limit of the gas detector tubes. This phenomenon was likely associated with the ability of urea to construct a highly polar and hydrogen-bond-rich micro-environment, which facilitated the dissolution and retention of H_2_S in the liquid phase. The incorporation of PP increased H_2_S removal across these systems, with the most pronounced synergistic effect being observed for B-U system, where the removal rate increased from 50.79% ± 2.75% to 72.22% ± 1.92%. This observation aligned with previous studies showing that deep eutectic solvents composed of basic hydrogen-bond acceptors enhanced H_2_S removal, owing to their selective binding affinity toward sulfur-containing gases [40]. Additionally, it is reported that phenolic-rich plant extracts can promote H_2_S uptake through physical adsorption mediated by their functional groups [41].

Overall, these results demonstrated that while NADESs exhibited intrinsic deodorization capabilities, the incorporation of PP further enhanced and stabilized the overall deodorization performance through synergistic interactions. NH_3_ capture was dominated by acid-base interactions reinforced by NADES-PP hydrogen-bond networks, whereas effective H_2_S removal relied more strongly on specific solvent-gas affinities and PP-assisted synergistic interactions, particularly in urea-based systems. Based on these mechanistic distinctions, CC-CA was selected for subsequent NH_3_-focused studies, while B-U was chosen for H_2_S removal investigations.

3.3. NADES Component Proportion Optimization for PP Extraction and Deodorization Performance

3.3.1. Optimization of CC-CA System for PP Extraction Under NH3 Removal Constraint

The CC-CA system was selected for further optimization based on its exceptional NH_3_ removal performance, achieving 100% removal even at 100-fold dilution in preliminary screening (Table 3). This complete removal created a ceiling effect that precluded the use of NH_3_ removal rate as a responsive optimization variable, as all CC-CA formulations would yield the same maximum value (100%). Therefore, a two-step strategy was adopted: primary optimization using TPC as the response variable to maximize polyphenol extraction yield, followed by post-optimization validation to confirm that the optimized extraction conditions preserved the exceptional NH_3_ removal performance.

Single-factor experiments were first conducted to examine the effects of key compositional and operational parameters on PP extraction (Figure 3). All three factors exhibited non-linear relationships with TPC, indicating that extraction capacity was governed by a balance between solvent structure and mass transfer behavior. The HBA:HBD molar ratio, expressed as CC:CA, showed a pronounced influence on TPC, which increased initially and then declined as the ratio increased, reaching a maximum at approximately 2:1. At this ratio, an appropriate proportion of citric acid enhanced hydrogen-bonding interactions between the NADES and phenolic hydroxyl groups, thereby improving PP solubility [42]. In contrast, excessive HBD content increased system viscosity and hindered mass transfer, resulting in reduced extraction capacity. A similar trend was observed for the H_2_O:HBD molar ratio. Optimal extraction was achieved at a ratio of 6:1, reflecting a typical dilution effect. Moderate water addition reduced viscosity and facilitated solute diffusion, whereas excessive water disrupted the hydrogen-bond network of the NADES, weakening its solvent capacity for PP [43]. L/S ratio also exhibited a clear optimum near 20:1, representing a balance between providing sufficient solvent for efficient mass transfer and avoiding unnecessary dilution of extracted phenols [44].

Based on the single-factor results, three critical factors, including HBA:HBD ratio (3, 2, and 1), H_2_O:HBD ratio (4, 6, and 8), and L/S ratio (20, 35, and 50), were selected as the center point for a BBD. The experimental design matrix and corresponding TPC values were summarized in Table 4. A quadratic polynomial model describing the response (TPC, Y) was established:

$$[eqn]$$

where A, B, and C represented the HBA:HBD ratio, H_2_O:HBD ratio, and L/S ratio, respectively. The model was highly significant (p < 0.001) with excellent goodness-of-fit (R^2^ = 0.9542, Table 5), indicating its suitability for describing the extraction behavior of PP in the CC-CA system. Response surface analysis (Figure 4) further revealed significant interactions among the variables, with the liquid-to-solid ratio (C) and its quadratic term (C^2^) exerting a particularly strong influence on TPC (p < 0.01). According to the model, the optimal extraction conditions were predicted as an HBA:HBD ratio of 2.242, an H_2_O:HBD ratio of 6.332, and a L/S ratio of 31.153:1, corresponding to a predicted TPC value of 141.485 mg/g. Experimental validation under these conditions yielded a TPC of 138.36 ± 0.22 mg/g, with a relative error of 2.21%, confirming the reliability of the BBD-RSM model.

Crucially, the CC-CA-PP complex prepared under the optimized conditions retained complete NH_3_ removal capability, achieving 100% removal even after 100-fold dilution. This result confirmed that the optimization strategy successfully enhanced PP extraction efficiency without compromising the intrinsic deodorization performance of the CC-CA system, thereby validating the feasibility of simultaneously optimizing extraction yield and functional performance.

3.3.2. Optimization of B-U System for PP Extraction as H2S Removal

In contrast to the CC-CA system, the optimization of B-U system focused exclusively on enhancing its H_2_S deodorization performance, as its PP extraction capability was relatively high compared to other NADES systems.

Employing H_2_S removal capacity as the evaluation metric, single-factor experiments were conducted to investigate the effects of the HBA:HBD (B:U) molar ratio and the H_2_O:HBD (H_2_O:U) molar ratio (Figure 5). The results revealed a nonlinear relationship between B:U ratio and removal capacity. As the ratio shifted from 2:1 to 1:3, H_2_S removal rate initially increased and then decreased, peaking at approximately 1:2. This trend suggested that a moderate increase in the urea proportion enhanced hydrogen-bonding interactions within the system, thereby facilitating H_2_S capture. However, excessive HBD content would elevate system viscosity, and impede gas diffusion, thus potentially leading to active site saturation, and collectively diminishing the removal capability [38]. Furthermore, the HBD:H_2_O molar ratio exerted a substantial influence on deodorization performance, with maximum H_2_S removal being observed at 1:6. This optimal ratio indicated that an appropriate water content reduced the solvent network structure and promoted mass transfer, whereas excessive water disrupted the stable hydrogen-bonding assembly of NADESs, thereby weakening effective interactions with H_2_S and compromising removal performance.

Therefore, the optimized B-U system ratio was determined as B:U:H_2_O = 1:2:12, which exhibited the best H_2_S removal performance within the experimental range.

3.4. Exploring the Deodorization Mechanism of NADES-PP Systems

To elucidate the acid-base behavior during gas absorption, the pH changes in CC-CA-PP/NH_3_ and B-U-PP/H_2_S systems were monitored over 30 min (Figure 6). Upon gas exposure, both systems showed a rapid initial pH shift followed by gradual stabilization. However, the direction, extent, and final equilibrium pH differed substantially, reflecting distinct gas–NADES interaction pathways.

The pristine CC-CA system was strongly acidic, with an initial pH of 2.49. Upon the introduction of NH_3_, the pH rose sharply within 0–2 min to 5.30, then declined slightly and stabilized around 4.94. This trend suggested that NH_3_ uptake was initially dominated by rapid acid-base neutralization. The coupled acid-base reactions and structural adjustments gradually reached a dynamic equilibrium, resulting in the observed pH stabilization. In contrast, the B-U-PP/H_2_S system exhibited a relatively high initial pH (pH = 8.51), followed by only a slight decrease upon H_2_S exposure, with rapid stabilization at around 8.20. This limited pH variation suggested that H_2_S underwent only weak acid-base neutralization under weak acid-weak base conditions. Therefore, H_2_S capture in the B-U-PP system was not driven by strong chemical neutralization, but was instead governed primarily by weak non-covalent interactions and physical dissolution [45].

Overall, the distinct pH evolution profiles reflect fundamental differences in interaction strength and capture pathways between NH_3_ and H_2_S in the respective NADES systems, which correlate with their contrasting deodorization capacities.

FT-IR analysis was employed as a qualitative tool to probe molecular interactions and elucidate the gas removal mechanisms in the NADES systems. As shown in Figure 7A, in the 3300–3000 cm^−1^ region, CC-CA exhibited a broad absorption band arising from overlapping O-H stretching vibrations associated with hydrogen bonding among citric acid carboxyl and hydroxyl groups within the system. These features reflected a highly interconnected hydrogen-bond network in the NADES. Upon the introduction of PP, this band broadened without the appearance of new characteristic peaks, indicating that the phenolic hydroxyl groups of PP participated in and modulated the pre-existing hydrogen-bond network rather than inducing new chemical reactions. In the 1720–1600 cm^−1^ region, CC-CA-PP showed no characteristic absorptions associated with carboxylate species, suggesting that PP addition altered the local hydrogen-bonding environment without causing deprotonation of citric acid. After exposure to NH_3_, the O-H stretching band in the 3300–3000 cm^−1^ region broadened and shifted toward lower wave numbers. Meanwhile, a new absorption band emerged near 1600 cm^−1^ accompanied by additional features in the 1450–1300 cm^−1^ region, which were attributable to $[eqn]$ -related modes [46]. These spectral features are consistent with an acid-base neutralization mechanism.

For H_2_S (Figure 7B), the B-U spectrum exhibited a broad N-H/O-H stretching band at 3500–3000 cm^−1^, reflecting the hydrogen-bond network formed between betaine and urea. The absorption at 1688 cm^−1^ is assigned to the C=O stretching vibration of urea, while the region around 1600 cm^−1^ corresponded to the anti-symmetric stretching of the betaine carboxylate group, overlapping with the urea C=O absorption. These features indicated that betaine and urea assembled into a stable supra-molecular structure through multi-point hydrogen bonding. Compared to B-U, the O-H absorption band of B-U-PP further broadened, while the positions of other characteristic peaks remained essentially unchanged, with no new absorption bands appearing. This suggested that the introduction of PP mainly reinforced and reorganized the existing hydrogen-bond network rather than inducing the formation of new chemical bonds [14]. Upon exposure to H_2_S, B-U-PP/H_2_S exhibited a slightly narrower O-H absorption band. No new characteristic absorption peaks appeared in the 1800–600 cm^−1^ fingerprint region, with only a slight shift and broadening being observed in the C-O/C-N-related vibrations near 1000 cm^−1^. These indicated that H_2_S incorporation primarily induced adjustments in the local electronic environment and hydrogen bond structure, but not new bond formation. Accordingly, H_2_S fixation in the B-U-PP system was dominated by non-covalent interactions.

FT-IR analysis revealed a clear mechanistic distinction between the two systems. NH_3_ capture in CC-CA-PP involved acid-base neutralization, whereas H_2_S uptake in B-U-PP was governed mainly by weak non-covalent interactions. This difference accounted for the observed selectivity in deodorization performance.

Because EGCG was both the main components of persimmon soluble tannin and the main sub-unit of persimmon insoluble tannin, we further used the molecular dynamics simulations to reveal the possible mechanisms of NH_3_ to be solvated within the NADES system with EGCG as model compound. Trajectory analysis showed that NH_3_ molecules did not establish direct contacts with EGCG, but were preferentially located within polar solvent domains composed of choline cations, citric acid, chloride ions, and water (Figure 8A). These domains constituted continuous, hydrogen-bond-rich regions capable of accommodating small polar molecules [47]. Hydrogen-bond statistics further clarified the nature of this solvation behavior (Figure 8B). Over the entire trajectory, the 30 NH_3_ molecules formed a total of 86 hydrogen bonds, indicating that each NH_3_ molecule typically participates in two to three hydrogen bonds. These interactions were dominated by water molecules, with citric acid providing additional hydrogen-bonding sites. This pattern suggested that NH_3_ was stabilized through incorporation into hydrogen-bond-rich solvent domains rather than through specific pairwise binding to individual components.

Although direct NH_3_-EGCG interactions were absent, the introduction of EGCG substantially increased the density of phenolic hydroxyl groups and enhanced the overall polarity of the solvent environment. This effect reinforced the continuity and robustness of the NADES–water hydrogen-bond network, thereby facilitating the accommodation and stabilization of NH_3_ within polar microdomains. Together with the experimental observation that deodorization capacity was improved upon EGCG addition, these findings suggested that EGCG primarily acted as a cooperative modulator of the hydrogen-bond network rather than as a direct binding partner. Consistently, root mean square deviation (RMSD) analysis showed that the CC-CA-EGCG/NH_3_ system exhibited a lower equilibrium RMSD than the corresponding CC-CA-EGCG system without NH_3_ (Figure 8D), indicating reduced conformational fluctuations upon NH_3_ incorporation. This decrease in RMSD suggested that the presence of NH_3_ promoted a more structurally stabilized configuration of the system. Such stabilization may arise from hydrogen-bond interactions between NH_3_ and solvent components, as well as from reduced molecular mobility within the solvent network [48]. Visualization of non-covalent interactions using amIGM further corroborated this interpretation. Continuous green isosurfaces surrounding NH_3_ were observed (Figure 8C), indicating attractive interactions dominated by hydrogen bonding and Van der Waals forces between NH_3_ and the NADES components as well as water molecules. The absence of red or blue isosurfaces suggested that neither strong steric repulsion nor covalent bond formation occurred. These electronic-structure features confirmed that the retention of NH_3_ was governed primarily by non-covalent interactions and therefore represented a physical binding process.

In contrast, no stable hydrogen-bond structures were detected between H_2_S and EGCG, the NADES constituents, or water in the corresponding simulation system. Instead, H_2_S preferentially resided in the relatively loose regions of the solvent matrix rather than within hydrogen-bond-dense domains. Hydrogen-bond analysis showed that the number of H_2_S-related hydrogen bonds remained essentially zero throughout the trajectory, indicating the lack of effective donor or acceptor sites for hydrogen-bond participation. The amIGM analysis further revealed only scattered pale-green or light-yellow isosurfaces around H_2_S, indicative of weak Van der Waals interactions and slight hydrophobic contacts, without signatures of strong attractive forces or new bond formation. Consistently, RMSD analysis showed that although the introduction of H_2_S slightly reduced the equilibrium RMSD of the B-U-EGCG system, the overall RMSD remained significantly higher than that of CC-CA-based systems. This limited stabilization effect mirrored the experimentally observed lower deodorization efficiency toward H_2_S and confirmed that hydrogen bonding was not the dominant mechanism for H_2_S capture in this system. Therefore, unlike the hydrogen-bond-assisted fixation observed for NH_3_, H_2_S removal was better described as weak physical adsorption combined with diffusion limitation. This mechanistic divergence provided a molecular-level explanation for the distinct selectivity of the two systems toward NH_3_ and H_2_S.

4. Conclusions

In summary, by screening an appropriate NADES system, not only PP could be extracted efficiently, but also the obtained NADES-PP mixtures showed potential capabilities on removal of NH_3_ and H_2_S. This study systematically elucidated the deodorization behavior and underlying mechanisms of a composite system based on PP and NADES toward NH_3_ and H_2_S. Our data demonstrated that the CC-CA-PP exhibited excellent removal capacity for NH_3_, whereas its effect on H_2_S was comparatively limited. Instead, the B-U-PP system was suitable for H_2_S removal. Mechanistic investigations integrating spectroscopic analysis, molecular dynamics simulations, and quantum chemical calculations revealed that NH_3_ removal was governed by acid-base neutralization, with the resulting ammonium species further being stabilized within the PP-regulated NADES hydrogen-bond network. NH_3_ participated as both a hydrogen-bond donor and acceptor, integrating into the hydrogen-bond network and achieving prolonged retention. In contrast, H_2_S only interacted weakly with the solvent network, mainly through Van der Waals and hydrophobic contacts, and behaved in a near-free-diffusion manner, indicating a fundamentally distinct removal pathway. Overall, these findings supported the proposed hypothesis that NADESs enhanced deodorization of PP through cooperative micro-environment regulation rather than irreversible chemical conversion. The regeneration and reusability of the proposed deodorants were not evaluated in the present study. Although the NADES-PP systems demonstrated effective deodorization performance, further investigation into their cyclic stability, regeneration efficiency, and long-term performance is necessary to fully establish their practical applicability. Future studies incorporating oligomeric and polymeric tannin models would further elucidate the impact of chain length and polymerization degree on gas-capture mechanisms, building upon the foundation established here.

Our work highlighted hydrogen-bond network engineering as a key strategy for achieving selective gas capture in bio-based solvent systems, and provided a molecular-level basis for the rational design of green, persimmon-derived deodorants. Future studies should focus on solvent recyclability and long-term performance to further expand practical applications.