A dual-mode γ-CD/Fe,N-CD nanozyme enabled by host–guest interactions for the dual-mode fluorometric and colorimetric monitoring of zearalenone in wheat flour

Abstract

Zearalenone (ZEN), a potent endocrine-disrupting mycotoxin produced by Fusarium fungi, frequently contaminates cereal-based foods worldwide, necessitating reliable analytical methods for food safety and regulatory compliance. Herein, we report an innovative dual-readout sensing strategy for both fluorometric and colorimetric quantification of ZEN, based on a unified catalytic nanoplatform of iron and nitrogen co-doped carbon dots decorated with γ-cyclodextrin (γ-CD/Fe, N-CDs). The enlarged macrocyclic cavity of γ-CD selectively accommodates the bulky 14-membered resorcylic acid lactone ring of ZEN via host–guest binding, simultaneously triggering two distinct signaling pathways: static quenching of the intrinsic carbon dot fluorescence through ground-state complex formation, and inhibition of TMB oxidation as the captured ZEN scavenges hydroxyl radicals generated by the proximal iron catalytic centers, reducing colorimetric signal intensity in a dose-dependent manner. The fluorescence channel exhibited linearity over 0.5–80.0 ng mL⁻¹ with an LOD of 0.21 ng mL⁻¹, while the colorimetric channel was linear over 5.0–55.0 ng mL⁻¹ with an LOD of 1.6 ng mL⁻¹. Applied to commercial wheat flour samples, the method yielded recoveries of 95.8–102.4% with RSDs below 4%, showing no significant deviation from conventional reference techniques. This bifunctional nanodevice offers a robust, mutually corroborating dual-channel platform for high-throughput ZEN surveillance in grain commodities.

Zearalenone (ZEN), frequently contaminates cereal-based foods worldwide, necessitating reliable analytical methods for food safety. Image created with the assistance of Google Gemini.

1. Introduction

Zearalenone (ZEN) is a potent non-steroidal mycotoxin originating primarily as a secondary metabolite from several Fusarium fungi, particularly Fusarium graminearum, Fusarium culmorum, and Fusarium roseum, which frequently infect crops around the globe.¹ From a chemical perspective, ZEN is classified as a resorcylic acid lactone, featuring a 14-membered macrocyclic lactone ring fused to a resorcinol (1,3-dihydroxybenzene) core.² This specific molecular architecture bestows ZEN with distinct optical characteristics, including absorption peaks near 236 nm and 274 nm, along with intrinsic fluorescence exhibiting excitation and emission maxima at roughly 274 nm and 460 nm, respectively.³ Agricultural commodities such as wheat, maize, barley, and various cereal-derived products are highly susceptible to ZEN contamination, which often co-occurs with other hazardous fungal toxins like deoxynivalenol, thereby compounding the overall risk to food safety.⁴ The primary toxicological threat posed by ZEN stems from its pronounced estrogenic activity; its structural resemblance to naturally occurring estrogens allows it to readily bind estrogen receptors, precipitating profound endocrine disruption, reproductive dysfunction, and infertility in livestock and potentially humans.⁵ In response to these severe physiological hazards, the European Food Safety Authority (EFSA) has delineated a tolerable daily intake (TDI) of 0.25 µg kg⁻¹ body weight per day.⁶ Furthermore, stringent maximum levels have been enforced by the European Union through Commission Regulation (EU) 2023/915, dictating thresholds of 100 µg kg⁻¹ for unprocessed cereals, 75 µg kg⁻¹ for maize-based products, and a rigorous 20 µg kg⁻¹ for infant foods.⁷ Consequently, establishing fast, ultrasensitive, and robust detection methodologies for quantifying ZEN in agricultural matrices is of paramount importance to ensure public health, uphold food safety standards, and satisfy strict legislative requirements.

Historically, the quantification of ZEN has relied heavily on conventional analytical instrumentation, including high-performance liquid chromatography coupled with fluorescence detection (HPLC-FLD),⁸ liquid chromatography-tandem mass spectrometry (LC-MS/MS),⁹ and gas chromatography-mass spectrometry (GC-MS).¹⁰ Additionally, antibody-based techniques such as enzyme-linked immunosorbent assays (ELISA)¹¹ and lateral flow assays (LFA) are frequently utilized. While chromatographic and mass spectrometric approaches deliver exceptional precision and limits of detection, their routine application is severely hindered by the need for prohibitive capital investments, rigorous operational expertise, and laborious sample preparation protocols—often demanding extensive organic solvent extraction and solid-phase extraction (SPE) cleanup—which collectively preclude their utility for point-of-care or field-deployable testing.^12,13 Conversely, while immunoassays like ELISA provide accelerated throughput, they remain plagued by inherent vulnerabilities, including substantial cross-reactivity with structurally related metabolites, inconsistent batch-to-batch antibody viability, strict thermal storage dependencies, and high production expenses.¹⁴ More recently, electrochemical and alternative biosensors have shown promise for ZEN screening;¹⁵ nevertheless, many of these systems struggle with narrow operational ranges, inadequate selectivity when challenged with complex biological matrices, and cumbersome sensor fabrication processes.¹⁶ These persistent analytical bottlenecks highlight the urgent demand for a straightforward, economical, highly sensitive, and selective sensing architecture that facilitates the rapid and dependable determination of ZEN directly within intricate cereal matrices, bypassing the need for exhaustive sample pretreatment.

In the realm of advanced nanomaterial-based mycotoxin detection, carbon dots (CDs) have garnered immense interest as versatile fluorescent nanoprobes due to their superb optical stability, adjustable photoluminescence, minimal biological toxicity, straightforward surface modifiability, and low-cost scalable synthesis from ubiquitous carbon sources.¹⁷ The strategic incorporation of heteroatoms, particularly nitrogen, into the carbon matrix significantly amplifies their quantum efficiency and enriches their surface functionality relative to undoped variants, positioning them as premier signal transducers in food safety monitoring.¹⁸ Simultaneously, metal-functionalized CDs have emerged as robust nanozyme candidates. Doping the carbon dot framework with transition metals like iron introduces dense Fenton-like catalytic active sites, capitalizing on the expansive surface area inherent to nanocarbon architectures.¹⁹ Consequently, Fe-doped CDs have been aggressively pursued for their intrinsic peroxidase-mimicking properties, whereby surface-coordinated iron facilitates the breakdown of H₂O₂ into highly reactive hydroxyl radicals (˙OH), which in turn oxidize chromogenic substrates such as 3,3′,5,5′-tetramethylbenzidine (TMB) to yield a naked-eye detectable color change without relying on fragile biological enzymes. To achieve target specificity, supramolecular host molecules such as cyclodextrins are routinely employed. For ZEN, the eight-membered γ-cyclodextrin (γ-CD) offers an optimal recognition cavity (∼8.3 Å) that perfectly accommodates the bulky 14-membered macrolide ring of the toxin via robust non-covalent inclusion.²⁰ Despite these isolated technological leaps, the integration of CD-based fluorescence with transition-metal nanozyme colorimetry within a singular framework remains sparse in mycotoxin analysis. More importantly, no existing research has successfully engineered γ-CD-decorated, iron and nitrogen co-doped carbon dots into a cohesive, dual-signal sensory platform specifically tailored for ZEN detection, representing a critical research void that this study directly resolves.²¹

Herein, we introduce a novel, highly sensitive, and exceptionally selective dual-mode analytical platform that synergistically couples turn-off fluorescence with colorimetric readouts for the swift quantification of ZEN in cereal-based food products. The foundation of this methodology relies on a uniquely engineered, multifunctional nanohybrid—γ-cyclodextrin-modified iron and nitrogen co-doped carbon dots (γ-CD/Fe, N-CDs). This singular nanomaterial operates concurrently as the target recognition moiety, the luminescent reporter, and the peroxidase-mimicking nanozyme, dictating that both detection modalities are triggered by a unitary molecular binding event. In the fluorescent regime, the highly specific host–guest encapsulation of ZEN within the γ-CD cavity sequesters the mycotoxin adjacent to the carbon dot core, precipitating static fluorescence quenching. In the parallel colorimetric regime, the intrinsic nanozyme activity of the γ-CD/Fe, N-CDs promotes the H₂O₂-mediated oxidation of TMB to produce a distinct blue hue. However, upon host–guest capture of ZEN on the nanoparticle periphery, the phenolic hydroxyl groups of the resorcinol moiety act as potent radical scavengers, intercepting the ˙OH generated at the iron active centers and thereby attenuating the TMB oxidation in a strictly concentration-dependent fashion. Crucially, because both signal modulations are fundamentally tethered to the exact same supramolecular recognition instance, the dual-channel outputs are mechanistically synchronized, ensuring unparalleled analytical accuracy and self-validation. The innovative merits of this architecture are threefold: (1) it deploys a single, all-in-one nanoplatform that harmonizes selective recognition, optical transduction, and catalytic signal amplification, circumventing the complexity of multi-component or intricate ratiometric systems; (2) it anchors dual distinct readouts to a solitary ZEN–γ-CD complexation process, providing internal cross-verification that drastically elevates diagnostic reliability; and (3) it successfully melds host–guest supramolecular chemistry with advanced Fe-nanozyme catalysis into an easily deployable, robust sensor that effortlessly navigates the complexities of real-world cereal matrices. Ultimately, this paradigm offers an elegant convergence of operational simplicity, theoretical transparency, and superior analytical resilience, successfully surmounting the principal deficiencies inherent to traditional ZEN detection paradigms.

2. Experimental

2.1. Chemical reagents and materials

Analytical grade standards of zearalenone, α-zearalenol, β-zearalenol, citrinin, ochratoxin A, and aflatoxin B1 were acquired from Sigma-Aldrich (St. Louis, MO, USA). LGC Standards (Teddington, UK) supplied patulin, T-2 toxin, 1-adamantanamine, fumonisin B1, and citric acid. Alternariol and penicillic acid were procured from Cayman Chemical (Ann Arbor, MI, USA). Various organic compounds including γ-cyclodextrin, ferulic acid, gallic acid, chlorogenic acid, and caffeic acid were obtained from Tokyo Chemical Industry (TCI, Tokyo, Japan). Cysteine, ascorbic acid, urea, quercetin, and catechin were sourced from Alfa Aesar (Thermo Fisher Scientific, Ward Hill, MA, USA). Maltose, starch, phytic acid, and glutathione were provided by Merck (Darmstadt, Germany). Inorganic salts including zinc chloride, copper sulfate, and ferric chloride were purchased from Honeywell Fluka (Charlotte, NC, USA), while magnesium chloride and calcium chloride were acquired from VWR Chemicals (Radnor, PA, USA). All chemicals possessed analytical purity or higher and were utilized as received unless specifically noted.

2.2. Apparatus and characterization instruments

Comprehensive specifications regarding the analytical instrumentation employed in this study can be found in the SI.

2.3. Synthesis of γ-CD/Fe,N-CDs

A facile, single-step hydrothermal strategy was implemented to fabricate iron- and nitrogen-co-doped carbon dots modified with γ-cyclodextrin (γ-CD/Fe,N-CDs). The synthetic procedure involved dissolving 2.0 g of citric acid, 1.0 g of urea, 0.2 g of iron(iii) chloride hexahydrate (FeCl₃. 6H₂O), and 0.6 g of γ-cyclodextrin in 20 mL of high-purity water. The mixture was continuously stirred until it became entirely homogenous. Subsequently, this precursor solution was placed into a 50 mL stainless-steel autoclave equipped with a Teflon liner. The sealed vessel underwent thermal treatment at 200 °C for a duration of 10 h. After cooling to ambient temperature naturally, a dark brown dispersion was obtained. This crude product was passed through a 0.22 µm filter membrane and subjected to dialysis (MWCO 1 kDa) against ultrapure water for 48 h, with the dialysate being refreshed frequently. Finally, the purified γ-CD/Fe,N-CDs were preserved in darkness at 4 °C for future applications.

2.4. Quantum yield determination

Detailed protocols and mathematical equations utilized to evaluate the quantum yield of the synthesized nanoparticles are presented in the SI.

2.5. Fluorescence-based detection of ZEN

To evaluate the fluorometric response, 50 µL of a 1 mg mL⁻¹ γ-CD/Fe,N-CDs stock dispersion was added to a 2.0 mL test tube. Following this, 100 µL of a known concentration of ZEN standard solution was introduced. The final volume was adjusted to 2.0 mL utilizing a 10 mM PBS solution (pH 7.0). The resulting mixture was left to react at ambient temperature for a period of 5 min. Upon completion of the incubation time, fluorescence emission profiles were acquired by employing an excitation wavelength of 470 nm, and the intensity at the maximum emission peak (580 nm) was recorded to quantify the analyte.

2.6. Colorimetric quantification of ZEN

The colorimetric measurements were performed within a 10 mM acetate buffer medium adjusted to pH 4.0, maintaining a total working volume of 2.0 mL. Initially, 100 µL of the prepared ZEN standard solutions at various concentrations were mixed with the buffer. This step was succeeded by the consecutive introduction of 50 µL of γ-CD/Fe,N-CDs (0.5 mg mL⁻¹), 100 µL of a 1.0 mM ethanolic TMB solution, and 50 µL of 10 mM hydrogen peroxide. The sample was briefly agitated using a vortex mixer and subsequently incubated in a water bath at 37 °C for 5 min. Following this reaction period, the absorption intensity at 652 nm was documented, employing a reagent blank as the reference.

2.7. Application to real sample analysis and matrix preparation

Commercial wheat flour samples were purchased from various supermarkets to serve as real matrices. Prior to extraction, the samples were pulverized thoroughly to yield a uniform, fine powder. A precise 5.0 g portion of each ground cereal was weighed into a 50 mL polypropylene centrifuge tube. For the extraction process, 25 mL of an acetonitrile-water mixture (84 : 16, v/v) was introduced. To ensure optimal recovery, the suspension was vigorously vortexed for 5 min and further ultrasonicated for 10 min. The separation of solid matter was achieved by centrifuging the mixture at 8000 rpm for 10 min under ambient conditions. The ensuing supernatant was recovered and filtered through a 0.22 µm nylon syringe filter. The filtrate was concentrated to near dryness under a gentle stream of nitrogen gas at 40 °C. The remaining dry residue was subsequently redissolved in either 10 mM PBS (pH 7.0) for the fluorescence assay or 10 mM acetate buffer (pH 4.0) for the colorimetric procedure, achieving a final volume of 1.0 mL. To mitigate potential matrix effects and guarantee that the analytical readings remained within the linear calibration ranges, all reconstituted solutions were diluted with their respective buffers prior to analysis.

To validate the reliability and accuracy of the proposed method, matrix blanks (cereal extracts free of ZEN) were artificially fortified with ZEN standards at three distinct concentration levels. These spiked samples were incubated for 10 min to allow complete equilibration before undergoing analysis. The quantification of ZEN via both the fluorometric and colorimetric strategies was executed exactly as detailed in Sections 2.5 and 2.6. The original analyte concentrations in the solid samples were determined by utilizing the constructed calibration plots, while adjusting for the initial sample mass and all subsequent dilution steps. Method recovery was expressed as the percentage of the quantified concentration relative to the initially spiked amount. Furthermore, the accuracy of the dual-mode methodology was cross-verified by analyzing selected cereal samples using liquid chromatography-tandem mass spectrometry (LC-MS/MS). Statistical parity between the two analytical approaches was evaluated by applying a paired Student's t-test, considering a p-value less than 0.05 as indicative of significant difference. Every analytical determination was conducted in triplicate, with the final data presented as the mean accompanied by the relative standard deviation (% RSD).

3. Results and discussion

3.1. Characterization of the prepared γ-CD/Fe, N-CDs

The morphological features of the synthesized γ-CD/Fe, N-CDs were examined by transmission electron microscopy (TEM). As shown in Fig. 1A, the nanoparticles are uniformly dispersed, quasi-spherical entities exhibiting no visible signs of aggregation, indicative of excellent stability in aqueous media. The particle size distribution histogram (Fig. 1A, inset) reveals a narrow diameter range spanning approximately 3 to 7 nm, with the predominant population centered around 5.9 nm.²² The X-ray diffraction (XRD) pattern (Fig. 1B) displays a single broad diffraction peak located near 2θ ≈ 23.43°, which corresponds to the (002) interlayer spacing of turbostratic graphitic carbon.²³ The absence of any sharp crystalline reflections confirms the predominantly amorphous character of the carbon dot matrix, a feature consistently observed in carbon nanomaterials derived through bottom-up hydrothermal carbonization of molecular precursors.

Fig. 1. (A) TEM image with particle size distribution histogram (inset), (B) XRD pattern, (C) FTIR spectrum, and (D) XPS survey spectrum of γ-CD/Fe, N-CDs.

The surface functional group composition was probed by Fourier-transform infrared (FTIR) spectroscopy (Fig. 1C). The broad absorption band spanning 3200–3450 cm⁻¹, with distinguishable contributions at 3433 cm⁻¹ and 3248 cm⁻¹, is ascribed to the overlapping stretching vibrations of O–H and N–H bonds, respectively, confirming the abundance of hydroxyl and amine functionalities on the nanoparticle surface.²⁴ A minor band at 2820 cm⁻¹ corresponds to aliphatic C–H stretching modes. The prominent and intense absorption at 1695 cm⁻¹ is assigned to C O stretching vibrations of carbonyl and carboxyl groups, while the band at 1411 cm⁻¹ is attributed to C–N stretching or aromatic C C skeletal vibrations.²⁵ The peak at 1254 cm⁻¹ arises from C–O stretching vibrations, and the bands observed at 835 cm⁻¹ and 706 cm⁻¹ correspond to out-of-plane bending deformations of aromatic C–H bonds. Collectively, these spectral features confirm the successful incorporation of nitrogen-containing moieties alongside oxygen-rich hydrophilic groups, endowing the carbon dots with excellent aqueous dispersibility.

The elemental composition and chemical bonding states of the γ-CD/Fe, N-CDs were further characterized by X-ray photoelectron spectroscopy (XPS). The survey spectrum (Fig. 1D) reveals three principal signals corresponding to C 1s (285.13 eV), N 1s (399.50 eV), O 1s (532.28 eV), and Fe (713.22 eV) confirming carbon, nitrogen, oxygen and iron as the major elemental constituents. The high-resolution C 1s spectrum (Fig. 2A) was deconvoluted into three component peaks at 284.13 eV, 284.99 eV, and 286.10 eV, which are attributed to sp²-hybridized graphitic carbon (C C/C–C), carbon–nitrogen bonds (C–N), and oxygenated carbon species (C O/C–O), respectively.²⁶ These assignments are fully consistent with the FTIR observations. The N 1s spectrum (Fig. 2B) was resolved into two peaks at 399.52 eV and 400.97 eV, corresponding to pyridinic nitrogen situated at the peripheral edges of the graphitic lattice and pyrrolic nitrogen embedded within five-membered heterocyclic rings, respectively.²⁷ The coexistence of both nitrogen configurations confirms the successful covalent integration of nitrogen dopants into the carbon framework. The O 1s spectrum (Fig. 2C) was deconvoluted into three peaks at 530.17 eV, 531.15 eV, and 532.54 eV, assignable to C O carbonyl groups, C–OH hydroxyl functionalities, and C–O–C ether linkages, respectively. The high-resolution Fe 2p₃/₂ spectrum exhibited contributions centered at approximately 709.7 and 711.8 eV, consistent with Fe²⁺ and Fe³⁺ species, respectively. Additional features at approximately 714.4 and 717.0 eV are attributable to multiplet-splitting and shake-up satellite structures characteristic of high-spin iron species (Fig. 2D), supporting the coexistence of mixed Fe²⁺/Fe³⁺ states in the γ-CD/Fe,N-CDs.

Fig. 2. High-resolution XPS spectra of (A) C 1s, (B) N 1s, (C) O 1s and (D) Fe 2p of the γ-CD/Fe, N-CDs. (E) UV-vis absorption, fluorescence excitation, and emission spectra of the γ-CD/Fe, N-CDs. (F) Excitation-dependent fluorescence emission spectra recorded at excitation wavelengths from 430 to 500 nm.

The optical characteristics of the prepared carbon dots were investigated by UV-vis absorption and photoluminescence spectroscopies (Fig. 2E). The UV-vis absorption spectrum (black dashed curve) exhibits a prominent absorption band centered at 283 nm, attributable to π–π* electronic transitions of the conjugated aromatic C C domains within the graphitic core, together with a broad secondary absorption extending around 440 nm, originating from n–π* transitions of surface-bound C O and C N chromophoric groups. The photoluminescence excitation spectrum (red curve) reveals a well-defined excitation maximum at 470 nm, and upon photoexcitation at this wavelength, the carbon dots emit an intense orange-red fluorescence band centered at 580 nm (blue curve). The large tokes shift of approximately 110 nm effectively minimizes self-absorption and reabsorption artifacts, which is highly favorable for fluorescence-based sensing applications. The excitation-dependent emission behavior was systematically investigated by recording fluorescence spectra at excitation wavelengths ranging from 430 to 500 nm (Fig. 2F). As the excitation wavelength was progressively increased, the emission intensity decreased gradually while the peak position exhibited a slight bathochromic shift, indicating a degree of excitation-dependent emission behavior. The maximum fluorescence emission intensity was achieved at an excitation wavelength of 470 nm, which was consequently adopted as the optimal excitation wavelength for all subsequent analytical measurements. The fluorescence quantum yield (QY) of the synthesized γ-CD/Fe,N-CDs was determined to be 16.5% using quinine sulfate as the reference standard.

3.2. Stability of the prepared γ-CD/Fe, N-CDs

For the proposed dual-mode sensing platform to be practically applicable in the analysis of complex cereal-based food matrices, the nanomaterial must maintain its fluorescence integrity under a broad range of environmental conditions. As illustrated in Fig. S1A, the fluorescence intensity of the γ-CD/Fe, N-CDs remained virtually constant across the entire investigated NaCl concentration range of 0.1 to 2.0 M, demonstrating outstanding tolerance to high ionic strength and confirming that increasing electrolyte content induces no detectable aggregation or surface charge screening of the nanoparticles. The influence of solution pH on fluorescence emission was evaluated over the range of pH 3 to 12 (Fig. S1B). The emission intensity was relatively low at strongly acidic conditions (pH 3–4), increased progressively through the weakly acidic range, and reached its maximum at pH 8, remaining essentially stable across pH 5 to 10.²⁸ A pronounced and progressive decline in fluorescence was observed only under strongly alkaline conditions, with emission intensity dropping markedly at pH 11 and decreasing further at pH 12, attributable to the deprotonation of surface hydroxyl and amine groups at extreme pH values, which disrupts the surface emissive states of the carbon dots. Given that pH 7.0 falls within the optimal stability window, it was selected as the working pH for the fluorometric assay. Photostability under prolonged UV irradiation was assessed over a 12-hour continuous illumination period (Fig. S1C). The emission intensity showed only a gradual and moderate decline after 12 hours of UV exposure, representing a reduction of less than 11%, which reflects commendable photostability markedly superior to that of conventional organic fluorophores. Finally, thermal stability was evaluated across the temperature range of 20 to 80 °C (Fig. S1D).²⁹ The fluorescence remained essentially unchanged from 20 °C up to 60 °C, with only a modest decrease recorded at 70 °C and 80 °C, corresponding to an overall attenuation of approximately 9% at the highest tested temperature. This slight reduction at elevated temperatures is attributed to thermally enhanced non-radiative relaxation pathways.³⁰ Collectively, these results confirm that the γ-CD/Fe, N-CDs exhibit robust fluorescence stability against variations in ionic strength, pH, prolonged UV illumination, and temperature, affirming their suitability for reliable analytical performance in real food sample testing.

3.3. Optimization of experimental conditions

To maximize the sensing performance of the γ-CD/Fe, N-CDs towards ZEN, vital assay parameters were systematically optimized. For the fluorometric detection channel, the buffer pH was evaluated between 4.0 and 10.0 (Fig. S2A). The most profound fluorescence quenching upon ZEN addition was recorded at pH 7.0, making it the preferred medium.³¹ At this neutral pH, the surface-anchored γ-CD maintains a structurally optimal macrocyclic cavity, facilitating exceptional host–guest interaction with ZEN, while the carbon dot's emissive states remain highly active.³² Conversely, extreme acidity prompted the protonation of functional groups, hampering the host–guest complexation, whereas strong alkalinity distorted the γ-CD cavity through deprotonation, diminishing the binding affinity. The interaction kinetics were studied by monitoring the incubation time from 1 to 30 minutes (Fig. S2B).^33,34 The quenching response stabilized within just 5 minutes, demonstrating that the ZEN molecule swiftly enters the γ-CD cavity to form a stable inclusion complex. Because the signal did not vary significantly after 5 minutes, this brief incubation period was adopted to ensure rapid assay turnaround.

For the colorimetric pathway, which capitalizes on the intrinsic peroxidase-like capabilities of the γ-CD/Fe, N-CDs to oxidize TMB, optimization targeted the pH, H₂O₂ concentration, temperature, and reaction duration. The buffer pH effect was mapped from 3.0 to 7.0 using acetate buffer (Fig. S3A). A maximum absorption signal at 652 nm was observed at pH 4.0, which perfectly aligns with the required mildly acidic environment for optimal Fenton-like decomposition of H₂O₂ into reactive ˙OH radicals by the iron catalytic centers.³⁵ As pH surpassed 4.0, catalytic efficiency dwindled, potentially due to the formation of iron hydroxides that block the active sites. Therefore, pH 4.0 was utilized for all subsequent colorimetric evaluations. The optimal H₂O₂ concentration was determined to be 10 mM from a tested range of 1–50 mM (Fig. S3B). This provided an adequate oxidant supply for robust chromogenic development without triggering non-enzymatic TMB oxidation, which typically occurs at overly high H₂O₂ levels and compromises assay sensitivity.³⁶ Temperature profiling revealed that 37 °C yielded the highest catalytic turnover without prematurely degrading the hydrogen peroxide (Fig. S3C), an issue prevalent at temperatures above 60 °C.³⁷ Lastly, the TMB oxidation process reached an absorbance plateau at 5 minutes (Fig. S3D), marking the completion of the chromogenic reaction and confirming the method's rapid response time.

3.4. Analytical performance of the dual-mode sensing platform for ZEN detection

Under the previously optimized experimental conditions, the quantitative capability of the bimodal γ-CD/Fe,N-CDs platform toward ZEN was systematically evaluated. Both the photoluminescence and absorptiometric channels demonstrated well-defined concentration-dependent responses across their respective analytical windows. In the fluorometric mode, a progressive and proportional attenuation of emission intensity was observed as the ZEN concentration increased from 0.5 to 80.0 ng mL⁻¹ (Fig. 3A). The resulting calibration curve (Fig. 3B) obeyed the linear regression equation F₀/F = 0.035C + 1.023, where F₀ and F represent the fluorescence intensity of the γ-CD/Fe,N-CDs in the absence and presence of ZEN, respectively, and C is the analyte concentration (ng mL⁻¹), yielding a coefficient of determination (R²) of 0.9965. In parallel, the colorimetric channel exhibited a clearly discernible, dose-dependent fading of the characteristic blue oxTMB absorption at 652 nm over the ZEN concentration range of 5.0 to 55.0 ng mL⁻¹ (Fig. 3C). The corresponding linear calibration (Fig. 3D) was described by ΔA = 0.0080C − 0.058, where ΔA denotes the change in absorbance at 652 nm, with R² = 0.9974.

Fig. 3. (A) Fluorescence emission spectra of γ-CD/Fe,N-CDs upon incremental ZEN addition, (B) corresponding fluorometric calibration plot over 0.5–80.0 ng mL⁻¹, (C) UV-vis absorption spectra illustrating the colorimetric response to ZEN, and (D) corresponding colorimetric calibration curve over 5.0–55.0 ng mL⁻¹.

The detection thresholds for both modalities were calculated according to the 3σ/S criterion, where σ is the standard deviation of ten replicate blank measurements and S is the slope of the respective calibration plot. The fluorometric channel afforded a limit of detection (LOD) of 0.21 ng mL⁻¹, while the colorimetric readout yielded an LOD of 1.6 ng mL⁻¹. The lower LOD achieved through the fluorescence pathway is attributable to the intrinsically superior signal-to-noise ratio characteristic of photoluminescence measurements, rendering this channel ideally suited for the precise quantification of sub-ng mL⁻¹ ZEN residues in cereal matrices. Concurrently, the colorimetric channel provides a rapid, instrument-free visual screening capability at moderate-to-high contamination levels. The integration of both channels within a single nanoplatform thus furnishes complementary analytical strengths — spanning from ultrasensitive instrumental quantification to convenient naked-eye assessment — that collectively enhance the practical versatility of the method.

The analytical performance of the proposed γ-CD/Fe,N-CDs dual-mode sensing platform was benchmarked against previously reported ZEN detection methods, as summarized in Table 1. The fluorometric mode achieved a linear range of 0.5–80.0 ng mL⁻¹ with a LOD of 0.21 ng mL⁻¹, while the complementary colorimetric mode operated over 5.0–55.0 ng mL⁻¹ with a LOD of 1.6 ng mL⁻¹. Compared to methods relying on molecularly imprinted polymers (MIPs) such as MIP/CdSe/ZnS QDs and CQDs-encapsulated MIP, which exhibited considerably higher LODs of 0.64 and 20 ng mL⁻¹, respectively, the proposed sensor demonstrates markedly superior sensitivity. Although certain specialized platforms — such as the evanescent wave optical-fiber aptasensor (LOD = 0.006 ng mL⁻¹) and the FLISA-based CdTe/CdS/ZnS QD system (LOD = 0.012 ng mL⁻¹) — achieve lower detection limits, these approaches require complex instrumentation, antibody conjugation, or sophisticated fabrication procedures, limiting their practical accessibility. Notably, a key distinguishing advantage of the present work is its dual-mode operation: the simultaneous fluorometric and colorimetric readout provides built-in cross-validation, significantly reducing false-positive results and enhancing reliability in complex food matrices such as wheat flour. Furthermore, the use of carbon-based nanomaterials avoids the toxicity concerns associated with heavy-metal QDs (CdTe, CdSe) employed in several competing methods. Collectively, the proposed sensor offers a favorable balance of sensitivity, selectivity, operational simplicity, and dual-signal verification, making it a practical and reliable tool for ZEN monitoring in food safety applications.

Table 1. Comparative evaluation of the proposed dual-mode method against previously reported ZEN detection approaches.

Probe/sensing strategy	Linear range (ng mL−1)	LOD (ng mL−1)	Matrix	Ref.
CdTe/CdS/ZnS QDs–anti-ZEN mAb (FLISA)	0.038–0.977	0.012	Maize	38
MIP on CdSe/ZnS QDs (fluorescence quenching)	0.96–993.3	0.64	Corn, rice, wheat flour	39
CQDs-encapsulated MIP (MIFQP, fluorescence quenching)	20–1000	20	Corn	40
N-CDs/AuNPs aptasensor (IFE, signal-on)	0.25–200	0.0875	Corn flour/corn oil	41
NCDs-aptamer/oxSWCNHs (fluorescence quenching)	20–100	18	Cereal products	42
FAM–aptamer/functional GO (turn-on)	0.5–64	0.5	Beer, wine	43
UCNPs–AuNRs aptasensor (FRET)	0.05–100	0.02	Wheat, maize	44
Evanescent wave optical-fiber aptasensor (Cy5.5, signal-off)	0.002–184.6	0.006	Corn flour	45
γ-CD/Fe,N-CDs (dual-mode) fluorometric	0.5–80.0	0.21	Wheat flour	This work
Colorimetric	5.0–55.0	1.6

3.5. Selectivity of the dual-mode sensing platform toward ZEN detection

To ascertain the specificity of the γ-CD/Fe,N-CDs toward ZEN, the sensor was challenged with an array of potential interfering substances typically found in agricultural commodities and food extracts (Fig. S4). The tested interferents included other prevalent mycotoxins (citrinin, ochratoxin A, aflatoxin B1, deoxynivalenol, fumonisin B1, T-2 toxin, patulin, alternariol, and penicillic acid), structurally related ZEN metabolites (α-zearalenol and β-zearalenol), common polyphenols (gallic acid, ferulic acid, caffeic acid, chlorogenic acid, catechin, quercetin), biological thiols and antioxidants (ascorbic acid, cysteine, glutathione), phytic acid, and various metal ions (Fe³⁺, Cu²⁺, Zn²⁺, Ca²⁺, Mg²⁺). Experimental findings confirmed that, barring ZEN, none of the bulk interferents caused substantial fluorescence quenching or disrupted the colorimetric readout. While the analogous metabolites α-zearalenol and β-zearalenol elicited a marginal cross-reactive response, their signal magnitude remained vastly inferior to that of ZEN, firmly establishing the platform's exceptional selectivity.

This robust discriminatory power relies on a synergistic dual-recognition mechanism defined by the distinct physicochemical properties of the target analyte. Primarily, the superior selectivity is dictated by the precise dimensional and structural complementarity between ZEN and the cyclodextrin host. The larger internal cavity of γ-CD (approximately 8.3 Å) is ideally suited to encapsulate the resorcinol moiety of ZEN, while its flexible macrolide chain wraps along the cavity rim, thereby establishing a remarkably stable inclusion complex. Other mycotoxins, such as aflatoxin B1 or deoxynivalenol, exhibit divergent molecular architectures, spatial volumes, or polarities that preclude favorable docking within the γ-CD cavity.⁴⁶ Similarly, native food polyphenols tend to be too bulky or rigid to form stable inclusion complexes, and inorganic ions lack the requisite hydrophobicity. In the secondary colorimetric mechanism, specificity is augmented by spatial proximity: only after ZEN is successfully sequestered by the γ-CD cavity is it brought close enough to the iron catalytic centers of the CDs. Once in this restricted microenvironment, the two phenolic –OH groups on the ZEN structure effectively scavenge the ˙OH radicals produced by the nanozyme, thereby impeding TMB oxidation and enabling highly specific colorimetric detection. Because both signal channels are governed by the identical host–guest binding event rather than through independent, uncorrelated pathways, a genuine ZEN signal must generate concordant attenuation in both channels simultaneously; any interferent capable of nonspecifically quenching fluorescence or reducing oxTMB via a mechanism unrelated to γ-CD occupancy would perturb only a single channel, furnishing an intrinsic cross-validation criterion that markedly reduces the probability of false-positive results in complex cereal matrices.

3.6. Mechanistic insights into the sensing process

To elucidate the fundamental pathways responsible for the observed fluorescence attenuation, the optical characteristics of ZEN were first evaluated. The UV-vis absorption profile of ZEN across the analytical concentration range was plotted alongside the photoluminescence excitation and emission spectra of the γ-CD/Fe, N-CDs (Fig. 4A). The functionalized carbon dots yielded a primary excitation maximum at roughly 470 nm and an emission peak at 580 nm. In contrast, ZEN exhibited prominent absorption bands situated around 236 nm, 274 nm and 315 nm. An examination of these spectral features revealed an absence of meaningful overlap between the optical absorption of the estrogenic mycotoxin and the emission or excitation signatures of the nanoprobes. Consequently, mechanisms heavily reliant on spectral intersection, namely Förster resonance energy transfer (FRET) and the inner filter effect (IFE), can be definitively ruled out as primary drivers of the signal reduction.

Fig. 4. (A) UV-vis absorption spectrum of ZEN and fluorescence excitation/emission spectra of γ-CD/Fe,N-CDs. (B) Fluorescence lifetime decay curves of γ-CD/Fe,N-CDs in the absence and presence of ZEN. (C) Fluorescence emission spectra of γ-CD/Fe,N-CDs and Fe, N-CDs with and without ZEN. (D) Competitive binding assay showing fluorescence recovery of γ-CD/Fe,N-CDs/ZEN upon addition of adamantane.

In order to differentiate between dynamic and static quenching modalities, the system's behavior was scrutinized at three distinct temperatures (298 K, 308 K, and 318 K) by applying the classical Stern–Völmer model (Fig. S5A):F₀/F = 1 + K_sv[Q]

Within this framework, F₀ and F denote the luminescence intensities of the γ-CD/Fe, N-CDs prior to and following ZEN introduction, respectively, while K_sv represents the quenching constant and [Q] denotes the analyte concentration. Across all evaluated temperatures, the resultant Stern–Völmer plots maintained excellent linearity, suggesting the predominance of a singular quenching pathway. The calculated K_sv values demonstrated a steady decline as the thermal energy of the system increased, dropping from 0.036 mL ng⁻¹ at 298 K to 0.033 mL ng⁻¹ at 308 K, and finally to 0.026 mL ng⁻¹ at 318 K. Such an inverse relationship with temperature strongly points toward static quenching. In this scenario, higher thermal energy progressively disrupts the non-fluorescent ground-state complex formed between ZEN and the γ-CD-modified surface, thereby lowering the association equilibrium and decreasing the overall quenching capability.

Verification of the static quenching hypothesis was achieved through time-resolved fluorescence decay analyses (Fig. 4B). Prior to the introduction of the target mycotoxin, the excited-state decay of the γ-CD/Fe, N-CDs fit well to a biexponential function, resulting in an average lifetime (τ₀) of 6.9 ns. When ZEN was added, the luminescence lifetime stayed remarkably constant, shifting marginally to 6.8 ns. This invariance in the excited-state duration confirms that the fluorophore's intrinsic decay kinetics are completely undisturbed by the mycotoxin, aligning perfectly with a ground-state complexation event.

Having confirmed static quenching as the underlying phenomenon, the stoichiometry and affinity of the interaction were quantified utilizing a modified logarithmic Stern–Völmer equation:log[(F₀ − F)/F] = log K_a + n log[Q]

At 298 K, the double-logarithmic plot of log[(F₀ − F)/F] versus log[ZEN] exhibited good linearity, with a regression coefficient of R² = 0.9923 (Fig. S5B). According to the relation log[(F₀ − F)/F] = log K_a + n log[Q], the slope of 0.934 corresponds to an apparent binding-site number (n) close to unity, indicating an interaction behavior consistent with approximately one dominant binding site. Because the original regression was constructed using ZEN concentration in ng mL⁻¹, the intercept cannot be directly interpreted as a conventional association constant in M⁻¹. After converting the ZEN concentration to molar units using a molecular weight of 318.36 g mol⁻¹, the transformed intercept corresponds to an apparent association constant of approximately 3.8 × 10⁶ M⁻¹. This relatively high apparent affinity supports a strong interaction between ZEN and the γ-CD/Fe,N-CDs recognition system.

To firmly establish that the enlarged macrocyclic cavity of γ-CD serves as the critical capture element, the fluorescence response of unmodified Fe, N-CDs was evaluated against the functionalized γ-CD/Fe, N-CDs under identical ZEN exposure (Fig. 4C). Introduction of 30 ng mL⁻¹ ZEN elicited a robust signal drop in the γ-CD-decorated dots, whereas the bare nanodots experienced only negligible dimming. This profound difference underscores the indispensable role of γ-CD-driven host–guest inclusion in mediating the sensing interaction.

Additional proof of this specific recognition mechanism was obtained via a competitive binding displacement assay using 1-adamantanamine, a well-known strong binder for the γ-CD cavity (Fig. 4D). When the γ-CD/Fe, N-CDs were pre-treated with 20 mM 1-adamantanamine, their subsequent quenching response to 30 ng mL⁻¹ ZEN was drastically impeded, falling to levels comparable to the background response of the bare carbon dots.

Alongside their fluorescent properties, the intrinsic nanozyme capabilities of the γ-CD/Fe, N-CDs were validated by tracking the catalytic oxidation of TMB at an absorbance wavelength of 652 nm (Fig. S5C). Control mixtures containing only TMB and H₂O₂ lacked any notable absorbance. Similarly, combining the nanodots with TMB in the absence of hydrogen peroxide yielded only a minimal background signal. However, the fully assembled system rapidly developed an intense blue coloration, verifying robust peroxidase-mimicking functionality.

To elucidate the reactive oxygen species driving the catalysis, electron paramagnetic resonance (EPR) spectroscopy was employed in conjunction with the spin-trapping agent 5,5-dimethyl-1-pyrroline N-oxide (DMPO) (Fig. S5D). The emergence of a classic 1 : 2 : 2 : 1 quadruplet signal conclusively identified hydroxyl radicals (˙OH) as the primary reactive intermediates.

To quantitatively characterize the catalytic efficiency of γ-CD/Fe,N-CDs and benchmark its performance against horseradish peroxidase and reported Fe-based nanozymes, steady-state kinetic analysis was performed using the Michaelis–Menten model (Fig. S6A–S6D).⁴⁷ Initial reaction velocities were measured at varying concentrations of one substrate while holding the other constant and saturating, allowing the apparent Michaelis constant (K_m) and maximum reaction velocity (V_max) to be determined independently for TMB and H₂O₂. For TMB as the variable substrate, γ-CD/Fe,N-CDs exhibited K_m = 0.10 mM and V_max = 3.00 × 10⁻⁸ M s⁻¹; for H₂O₂ as the variable substrate,⁴⁷K_m = 0.80 mM and V_max = 3.00 × 10⁻⁸ M s⁻¹ were obtained, both confirmed by linear Lineweaver–Burk transformation of the same dataset. The essentially identical V_max values obtained independently from the two substrate titrations reflect the same maximum turnover rate of the same catalytic Fe active sites, providing an internal consistency check on the kinetic analysis. The K_m value for TMB was substantially lower than that reported for native HRP (0.40 mM) and comparable Fe-MOF nanozymes (0.31 mM), indicating markedly higher substrate affinity of γ-CD/Fe,N-CDs toward TMB relative to these reference catalysts.⁴⁸ This enhanced catalytic performance is attributed to the high density of accessible Fe active sites within the carbon dot framework and improved substrate accessibility afforded by the hydrophilic surface environment provided by γ-CD and nitrogen-containing functional groups.

Collectively, the mechanistic investigation establishes a coherent and fully experimentally substantiated dual-mode sensing mechanism for the γ-CD/Fe,N-CDs platform. In the fluorescence channel, ZEN undergoes specific host–guest inclusion complexation within the γ-CD cavity anchored on the Fe,N-CD surface. The resulting ground-state inclusion complex quenches the nanocarbon fluorescence through a purely static mechanism that is completely independent of spectral overlap. In the colorimetric channel, the Fe active sites of the same γ-CD/Fe,N-CDs nanomaterial catalyze Fenton-type H₂O₂ decomposition to generate ˙OH radicals that oxidize TMB to the blue oxTMB product; ZEN bound within the γ-CD cavity in close proximity to these Fe active sites scavenges ˙OH radicals with enhanced local efficiency, causing concentration-dependent inhibition of TMB oxidation and progressive bleaching of the colorimetric signal. Crucially, both signal channels are driven by the same singular host–guest recognition event, providing mechanistic coherence, mutual analytical corroboration, and substantially enhanced detection reliability.

3.7. Practical execution in authentic cereal matrices

The practical applicability of the dual-signal analytical platform was evaluated through recovery experiments conducted on commercially available cereal-based food product (wheat flour) artificially fortified with known quantities of ZEN. Results were cross-validated against a standardized LC-MS/MS protocol serving as an independent reference method (Table 2).⁴⁹ For the fluorescence-based readout, cereal extracts were spiked at three concentration levels —0.5, 5.0, and 10.0 ng mL⁻¹. The proposed method delivered robust recovery rates ranging from 95.8% to 102.4%, accompanied by relative standard deviations (RSDs) between 2.36% and 3.95%, closely matching the corresponding LC-MS/MS recoveries of 97.1–101.6% (RSD 2.14–3.48%). For the colorimetric channel, fortification levels were set at 5.0, 15.0, and 25.0 ng mL⁻¹. This pathway yielded satisfactory recoveries of 96.2% to 100.8% with RSDs spanning 2.53% to 3.72%, in close accordance with LC-MS/MS recoveries of 97.5–101.9% (RSD 2.27–3.41%). Across both detection channels, recoveries consistently fell within the 95–103% range at all spiking levels, with RSD values uniformly below 4.0%, confirming excellent accuracy and inter-assay reproducibility that comfortably satisfy the internationally accepted performance criteria for mycotoxin quantification in complex food matrices.

Table 2. Comparison of the proposed fluorometric and colorimetric methods with LC-MS/MS for the determination of ZEN in spiked wheat flour samples.

	Proposed method				LC-MS/MS
Method	Added (ng mL−1)	Found (ng mL−1)	Recovery (%)	RSD (%)	Found (ng mL−1)	Recovery (%)	RSD (%)	t-test*
Fluorometric	0	N.D.	—	—	N.D.	—	—	1.27
	0.5	0.479	95.8	3.95	0.491	98.2	3.48
	5.0	5.12	102.4	2.36	5.08	101.6	2.14
	10.0	9.84	98.4	2.73	9.71	97.1	2.86
Colorimetric	0	N.D.	—	—	N.D.	—	—	1.36
	5.0	5.04	100.8	2.53	4.88	97.5	3.41
	15.0	14.43	96.2	3.72	15.29	101.9	2.27
	25.0	24.65	98.6	3.18	25.18	100.7	2.85

The statistical agreement between the proposed methodology and the reference LC-MS/MS method was evaluated using a paired Student's t-test based on the three matched concentration levels for each detection channel. The calculated t-values were 1.27 for the fluorometric method and 1.36 for the colorimetric method. Both values were lower than the two-tailed critical t-value of 4.303 at the 95% confidence level (df = 2, n = 3 paired observations), indicating that no statistically significant difference was detected between the proposed methods and LC-MS/MS. Together with the satisfactory recovery values and low RSDs, these results demonstrate good agreement between the proposed dual-mode platform and the reference chromatographic method. The proposed approach additionally offers simpler instrumentation and a shorter analytical procedure than LC-MS/MS.

To further validate the real-world utility of the platform, six retail wheat flour (designated S1 through S6), were procured from local markets and screened for naturally occurring ZEN contamination without artificial spiking. Results were independently verified by LC-MS/MS analysis (Table S1). ZEN was positively detected in four of the six samples, with concentrations determined by the fluorometric channel ranging from 13.2 to 74.8 µg kg⁻¹ and those obtained via the colorimetric pathway spanning 12.4 to 74.5 µg kg⁻¹, while two samples (S3 and S5) returned ZEN levels below the limit of detection across all three analytical methods. The concentrations quantified by both proposed channels demonstrated excellent mutual agreement and correlated closely with the corresponding LC-MS/MS values across all detectable samples. Importantly, all positive detections remained compliant with the stringent European Union maximum residue limits (75 µg kg⁻¹). These findings collectively affirm the reliability, accuracy, and practical robustness of the proposed dual-mode γ-CD/Fe, N-CDs platform for the direct screening and quantification of naturally occurring ZEN contamination in commercial cereal food commodities.

The distinctive contribution of the γ-CD/Fe,N-CDs platform lies in integrating host–guest recognition, fluorescence reporting, and peroxidase-like catalysis within a single nanomaterial. Previous cyclodextrin-based approaches exploited inclusion-induced enhancement of ZEN's native fluorescence,^31,32 whereas the present system couples γ-CD-mediated ZEN recognition with quenching of carbon-dot fluorescence and inhibition of TMB oxidation. Compared with representative antibody- or aptamer-based fluorescence/colorimetric assays,^50,51 this design avoids biological recognition reagents and integrates the fluorescent and catalytic functions without assembling separate nanoparticle reporters. The two readouts provide complementary measurements, with fluorescence offering greater sensitivity and colorimetry permitting absorbance-based quantification. Application to wheat flour yielded recoveries of 95.8–102.4% with RSDs below 4%, supporting its analytical applicability in this matrix. Nevertheless, the fluorescence and colorimetric detection limits of 0.21 and 1.6 ng mL⁻¹, respectively, do not surpass those of several existing dual-mode sensors. The channels also require separate assay conditions, and sample extraction and concentration remain necessary. Moreover, cyclodextrin binding does not inherently ensure discrimination against all structurally related ZEN metabolites, while redox-active matrix constituents may affect the catalytic response; agreement between the channels therefore cannot exclude shared interference. Thus, the principal advance is the integration of supramolecular recognition and complementary optical transduction in an antibody- and aptamer-free platform, rather than the individual sensing principles or universally superior analytical performance.

4. Conclusion

In summary, a facile and highly dependable bimodal sensory architecture has been successfully engineered for the trace quantification of zearalenone, a potent endocrine-disrupting mycotoxin, in cereal-based commodities. This strategy leverages the unique properties of a unified nanoscale material: γ-cyclodextrin-decorated iron and nitrogen co-doped carbon dots (γ-CD/Fe,N-CDs). The surface-grafted γ-CD macrocycles furnish a precisely sized cavity that facilitates the selective entrapment of ZEN. This localized host–guest interaction concurrently triggers two distinct readouts: it extinguishes the intrinsic photoluminescence of the carbon dots via static ground-state complexation, while simultaneously dampening the material's peroxidase-mimetic activity through the targeted scavenging of hydroxyl radicals near the iron catalytic centers. Because both the optical and visual signals stem from a singular molecular recognition event, they offer cross-verifiable data, substantially elevating the analytical confidence compared to conventional solitary-mode sensors. The integrated platform exhibited outstanding linear range, sensitivity, and resistance to matrix interference. Its effective deployment in screening complex maize and wheat matrices, corroborated against standardized chromatographic protocols, underscores its immense potential for routine agricultural and food safety monitoring. Broadly, this research introduces a mechanistically sound, economical, and field-adaptable paradigm for tracing estrogenic contaminants, which could foreseeably be expanded to profile a wide array of small-molecule hazards in agricultural matrices.

Conflicts of interest

There are no conflicts to declare.

Data availability

Data will be available on request.

Supplementary information (SI) is available. See DOI: https://doi.org/10.1039/d6ra07703j.