A bundle of carbon fiber tow (approx. 1 cm length, 0.5 cm width) was sonicated sequentially in acetone, ethanol, and DI water for 15 min each to remove surface contaminants. To activate the surface and introduce oxygen-containing functional groups, the cleaned CF was immersed in 6.0 M HNO₃ at 80°C for 6 h [18]. The acid-treated CF was then washed thoroughly with DI water until the pH was neutral and dried in an oven at 60°C. Prior to use or modification, the activated CF was further electrochemically cleaned by cyclic voltammetry (CV) in 0.5 M H₂SO₄, sweeping the potential between −0.5 V and +1.5 V at 100 mV/s until a stable voltammogram was achieved. This electrode is designated as Bare CF.

For comparative purposes, Bi₂S₃ nanorod powder was synthesized via a solvothermal method [19]. In a typical procedure, 1.0 mmol of Bi(NO₃)₃·5H₂O and 3.0 mmol of thiourea were dissolved in 40 mL of ethylene glycol (EG) under vigorous stirring. The homogeneous solution was transferred into a 50 mL Teflon-lined stainless-steel autoclave, sealed, and maintained at 180°C for 12 h. After cooling, the black precipitate was collected by centrifugation, washed alternately with absolute ethanol and DI water three times, and finally dried at 60°C overnight. This powder is designated as Bi₂S₃-NRs.

Bi₂S₃/CF (Drop-cast Control) Electrode: A stable catalyst ink was prepared by dispersing 5.0 mg of the synthesized Bi₂S₃-NRs powder into 1.0 mL of a solution containing 950 μL of ethanol and 50 μL of 5 wt% Nafion solution. The mixture was ultrasonicated for 30 min to form a homogeneous black suspension. The Bi₂S₃/CF control electrode was fabricated by drop-casting 10 μL of this ink onto the surface of the Bare CF electrode and allowing it to dry at room temperature.

Bi₂S₃-CF (In-situ Hybrid) Electrode: The proposed hybrid electrode was fabricated using a one-pot, in-situ hydrothermal growth method [20]. A piece of the pretreated Bare CF (approx. 3 cm in length) was fixed in a 50 mL Teflon-lined stainless-steel autoclave. In this process, 1.0 mmol of Bi(NO₃)₃·5H₂O and 3.0 mmol of thiourea were dissolved in 40 mL of DI water (not EG) under stirring. The Bare CF was fully immersed in this precursor solution. The autoclave was sealed and heated in an oven at 160°C for 10 h. During this process, the thiourea decomposes to provide S²⁻ ions, which react with Bi³⁺ ions, allowing Bi₂S₃ nanorods to nucleate and grow directly on the activated sites of the CF surface. After cooling, the Bi₂S₃-CF hybrid electrode was removed, rinsed thoroughly with DI water, and dried at 60°C.

The electrochemical properties of the electrode-electrolyte interface were characterized using CV and Electrochemical Impedance Spectroscopy (EIS) in a 0.1 M KCl solution containing 5.0 mM [Fe(CN)₆]^3−/4− as a redox probe [21]. CVs were typically recorded from −0.2 V to +0.6 V at a scan rate of 50 mV/s. EIS measurements were performed at the formal potential of the probe (+0.18 V) with a sinusoidal amplitude of 5 mV over a frequency range from 100 kHz to 0.1 Hz.

The electrochemical detection of ZEA was performed using Differential Pulse Voltammetry (DPV) [22]. The supporting electrolyte for all ZEA measurements was 0.1 M phosphate-buffered saline (PBS). Experimental parameters were optimized, including the pH of the PBS (from 4.0 to 8.0) and the accumulation time (from 0 s to 300 s) under open-circuit potential. The standard DPV parameters used were: modulation amplitude 0.05 V, modulation time 0.05 s, and step potential 0.005 V. All experiments were conducted at room temperature. For LOD estimation, the blank was defined as 0.1 M PBS (pH 6.0) containing 0 ng/mL ZEA and measured under identical accumulation and DPV conditions; σ was calculated as the standard deviation of the baseline-corrected blank current recorded at the ZEA peak potential from repeated blank measurements, and the LOD was obtained using LOD = 3σ/slope in accordance with common analytical-chemistry practice.

Cereal samples (corn flour, wheat flour, and oat flakes) were purchased from a local supermarket. The absence of ZEA in these samples was first confirmed by a standard HPLC-MS/MS method. For spike-and-recovery tests, ZEA extraction was performed based on established protocols. Briefly, 5.0 g of homogenized cereal sample was accurately weighed and mixed with 20 mL of an acetonitrile-water solution (80:20, v/v). The mixture was vortexed for 30 min and then centrifuged at 5000 rpm for 10 min. The supernatant was collected, and 1.0 mL was transferred to a new tube and evaporated to dryness under a gentle N₂ stream. The residue was then redissolved in 1.0 mL of 0.1 M PBS (pH 6.0) to create the final sample extract. Quantification in all cereal extracts was performed by the standard addition method to compensate for matrix effects, as recommended for calibration in complex matrices where matrix-matched standards are not readily available. Briefly, an aliquot of the sample extract was placed in the electrochemical cell and the DPV peak current (I) of the extract was first recorded (unspiked). Subsequently, small aliquots of ZEA working standard were added stepwise to the same extract, and DPV was recorded after each addition under identical conditions. A standard addition plot was constructed by linear regression of I versus the added ZEA concentration (Cadd). The ZEA concentration in the original extract (Csample) was determined from the x-intercept of the regression (Csample = −intercept/slope). All standard-addition determinations were performed in triplicate to obtain mean values and RSD.

The core objective of this study was to fabricate a hybrid electrode where the catalyst (Bi₂S₃) and the conductive substrate (CF) are seamlessly integrated. This was achieved via an in-situ hydrothermal growth strategy, designed to leverage the activated functional groups on the CF surface as nucleation anchors for the Bi₂S₃ nanorods. To validate this approach, the morphology of the in-situ Bi₂S₃-CF hybrid was meticulously compared against control samples: the Bare CF substrate and the Bi₂S₃/CF electrode (fabricated by drop-casting pre-synthesized Bi₂S₃-NRs powder).

Fig. 1 presents the SEM morphological analysis of the materials. Fig. 1a shows the Bare CF substrate, revealing bundles of smooth, cylindrical fibers, each with a diameter of approximately 7 μm. After acid and electrochemical activation (Fig. 1b), the fiber surfaces appear visibly roughened and etched, a process known to increase the physical surface area and expose reactive C–O and C=O functional groups. Fig. 1c displays the morphology of the control Bi₂S₃/CF electrode. As hypothesized, the drop-casting method results in a highly non-uniform and poorly structured surface. Large, irregular agglomerates of the Bi₂S₃ powder are seen physically resting on the CF surface, held in place by the Nafion binder. Large areas of the CF substrate remain exposed, while other areas are masked by thick, insulating clumps, which would invariably lead to poor catalytic efficiency and high contact resistance.

In stark contrast, Fig. 1d reveals the morphology of the Bi₂S₃-CF hybrid. The in-situ growth method has produced a radically different structure. The carbon fibers are completely and uniformly coated with a dense, 3D hierarchical array of interpenetrating nanorods [23]. These nanorods grow radially outwards from the CF surface, creating a “bristle-brush” or “nanorod-on-fiber” architecture that is highly ordered and porous. This structure prevents the agglomeration seen in the control sample and creates a vastly more extensive and accessible surface. This 3D architecture is anticipated to significantly enhance mass transport of the ZEA analyte to the active sites and provide a multitude of accessible catalytic surfaces [24].

To investigate the individual nanorods in more detail, TEM analysis was performed on the Bi₂S₃ material scraped from the hybrid electrode (Fig. 2). Fig. 2a (TEM) confirms the material is composed of discrete nanorods with uniform dimensions, averaging 20–30 nm in diameter and 150–200 nm in length. The HRTEM image in Fig. 2b provides a high-magnification view of a single nanorod, revealing highly crystalline, continuous, and clear lattice fringes. The measured inter-planar spacing of 0.39 nm can be indexed to the (220) plane of orthorhombic Bi₂S₃, confirming the high crystallinity of the nanorods along this growth axis. The corresponding SAED pattern (Fig. 2c) displays a set of bright, concentric, and dotted rings. These rings can be indexed to the (130), (220), (211), and (002) crystal planes of the orthorhombic Bi₂S₃ phase. The dotted-ring pattern confirms the nanocrystalline nature of the material, consistent with its formation from an assembly of nanorods. This high-quality, crystalline nanostructure is essential for efficient electrocatalysis.

The crystal phase and structural integrity of the synthesized materials were investigated using XRD and Raman spectroscopy. Fig. 3a shows the XRD patterns for the Bare CF, the Bi₂S₃-NRs powder, and the Bi₂S₃-CF hybrid. The diffractogram for Bare CF exhibits a single, broad diffraction peak centered at a 2θ value of approximately 25.5°, which is characteristic of the (002) plane of amorphous or graphitic carbon [23]. The Bi₂S₃-NRs powder displays a series of sharp, well-defined diffraction peaks at 2θ values of 22.4°, 24.9°, 28.7°, 31.8°, and 45.4°. These peaks are perfectly indexed to the (130), (211), (221), (131), and (431) planes, respectively, of the orthorhombic bismuthinite phase of Bi₂S₃ (JCPDS Card No. 17-0320) [25]. The absence of any other peaks confirms the high phase purity of the solvothermally synthesized nanorods. The broadening of the peaks, when analyzed with the Scherrer equation, indicates an average crystallite size of approximately 25 nm, which is in excellent agreement with the TEM observations. The XRD pattern for the Bi₂S₃-CF hybrid shows the characteristic sharp peaks of orthorhombic Bi₂S₃ superimposed on the broad (002) carbon peak. This co-existence definitively proves that the Bi₂S₃ nanorods were successfully synthesized and anchored onto the carbon fiber substrate in-situ and that the hydrothermal process did not alter the crystal phase of either component.

Raman spectroscopy was employed to provide complementary structural information, particularly concerning the carbon-chalcogenide interface (Fig. 3b). The Bare CF spectrum displays the two characteristic peaks for carbonaceous materials: the D-band at ∼1333 cm⁻¹, which corresponds to disordered carbon and structural defects, and the G-band at ∼1560 cm⁻¹, attributed to the in-plane sp² vibrations of graphitic carbon [26]. The Bi₂S₃-NRs powder shows several characteristic low-frequency phonon modes, including A_g modes at ∼186 cm⁻¹ and ∼239 cm⁻¹, and a B_1g mode at ∼266 cm⁻¹, which are in good agreement with literature values for Bi₂S₃ nanostructures [17]. The spectrum for the Bi₂S₃-CF hybrid clearly shows both sets of peaks: the D and G bands from the CF substrate and the characteristic Bi₂S₃ phonon modes. The intensity ratio of the D and G bands (I_D/I_G) provides insight into the defect density of the carbon material. The I_D/I_G ratio for the Bi₂S₃-CF hybrid was calculated to be 1.15, which is noticeably higher than that of the Bare CF (1.02). This increase suggests that the high-temperature, high-pressure hydrothermal process introduced additional defects or functional groups onto the CF surface. It is precisely these defects that are believed to act as the preferential nucleation sites for the Bi₂S₃ nanorod growth, supporting the proposed in-situ mechanism.

XPS analysis was conducted to verify the surface elemental composition and chemical valence states of the Bi₂S₃-CF hybrid material (Fig. 4). The survey spectrum (Fig. 4a) confirmed the presence of Bi, S, C, and O, with no other elemental impurities detected. The high-resolution Bi 4f spectrum (Fig. 4b) was deconvoluted into two sharp, symmetric peaks at binding energies of and ∼158.5 eV (Bi 4f_7/2) [27]. The spin-orbit splitting of 5.3 eV is a characteristic signature of the Bi³⁺ oxidation state. The high-resolution S 2p spectrum (Fig. 4b) was deconvoluted into a primary doublet with peaks at ∼160.9 eV (S 2p_3/2) and ∼162.1 eV (S 2p_1/2), which are characteristic of the S²⁻ (sulfide) state in metal sulfides. A minor peak at a higher binding energy of ∼168.8 eV can be attributed to a small amount of surface oxidation (S-O bonds), which is common for nanoscale sulfides exposed to air. The high-resolution C 1s spectrum (Fig. 4c) provided critical evidence for the hybrid’s interfacial chemistry. The spectrum was deconvoluted into three main components: a strong peak at ∼284.6 eV corresponding to C-C/C=C bonds from the graphitic fiber backbone, a peak at ∼286.0 eV (C-O, alcohols/ethers), and a peak at ∼288.5 eV (O-C=O, carboxylic acids). The presence of these oxygen-containing functional groups, introduced during the HNO₃ activation step, confirms the existence of chemical anchors on the CF surface. These groups facilitate the in-situ nucleation and growth of the Bi₂S₃ nanorods, leading to the strong, chemically-bonded interface hypothesized.

The physical surface properties, which are critical for sensing, were quantified using N₂ adsorption-desorption analysis. Table 1 provides the quantitative summary of these surface properties and provides the first definitive proof of the “Structural Synergy.” The Bare CF possesses a low BET specific surface area (SSA) of 15.2 m²/g. The pre-synthesized Bi₂S₃-NRs powder has a moderate SSA of 45.8 m²/g. Notably, the drop-cast Bi₂S₃/CF control electrode has an even lower SSA (28.3 m²/g) than the powder alone. This is attributed to the Nafion binder and the particle agglomeration, which blocks pores and reduces the accessible surface. The most significant finding is the SSA of the Bi₂S₃-CF hybrid: 120.5 m²/g. This value is 7.9-fold higher than the Bare CF and 2.6-fold higher than the Bi₂S₃ powder. This demonstrates a powerful structural synergy, where the CF acts as a 3D scaffold that prevents the nanorod agglomeration, and the in-situ grown nanorods, in turn, create a high-surface-area, porous network. This massive increase in SSA is a primary contributor to the enhanced sensing performance [28].

To test the hypothesis of “Electronic Synergy,” the electron transfer capabilities of the different electrodes were evaluated. This was done using the [Fe(CN)₆]^3−/4− redox probe, which provides a standard measure of interfacial kinetics. Fig. 5a presents the comparative CVs. The Bare CF electrode exhibits broad, sluggish redox peaks with a large peak-to-peak separation (ΔE_p) of ∼248 mV, indicating very poor electron transfer kinetics. The drop-cast Bi₂S₃/CF control shows a slight improvement, with higher peak currents (I_p) but a still-large ΔE_p of ∼172 mV. In dramatic contrast, the Bi₂S₃-CF hybrid displays a pair of sharp, well-defined, and quasi-reversible redox peaks. The peak currents are significantly larger, and the ΔE_p is only ∼65 mV. This value is very close to the theoretical value of 59 mV for an ideal one-electron reversible process, proving that the in-situ hybrid possesses vastly superior electron transfer kinetics [29].

EIS was used to quantify this kinetic difference by measuring the charge transfer resistance (R_ct), which is represented by the diameter of the semicircle in the Nyquist plot (Fig. 5b). The Bare CF shows a massive semicircle, with a calculated R_ct of ∼2610 Ω, confirming its high resistance to electron transfer. The Bi₂S₃/CF control shows a smaller semicircle, with R_ct ≈ 940 Ω. The high resistance is likely due to the poor physical contact between the agglomerated powder and the CF, as well as the insulating properties of the Nafion binder. The Bi₂S₃-CF hybrid exhibits a negligible semicircle that is almost a straight line, corresponding to an exceptionally low R_ct ∼75 Ω. This 35-fold decrease in R_ct compared to Bare CF, and 12-fold decrease compared to the drop-cast control, is unequivocal quantitative proof of the electronic synergy. The in-situ growth creates an intimate, chemically-bonded, and low-resistance electronic interface that allows for rapid electron shuttling between the CF substrate and the Bi₂S₃ catalytic sites. This low R_ct is a direct prerequisite for a highly sensitive sensor [30].

With the superior material properties of the Bi₂S₃-CF hybrid established, we proceeded to optimize its electrochemical response to ZEA. ZEA contains a phenolic hydroxyl group, which can be electrochemically oxidized. First, the effect of the supporting electrolyte pH on the DPV signal was investigated for 100 ng/mL ZEA (Fig. 6a). As shown in the plot of peak current versus pH (Fig. 6b), the oxidation peak current increases from pH 4.0 to 6.0, and then decreases as the solution becomes more basic. This bell-shaped curve is typical for phenolic compounds. At low pH, the phenolic group is fully protonated and more difficult to oxidize. As pH increases, deprotonation is facilitated, leading to an easier oxidation. However, at pH > 6.0, the ZEA molecule (and potentially the sensor surface) may become negatively charged, causing electrostatic repulsion and a decrease in signal. Therefore, a pH of 6.0 was selected as the optimal condition for all subsequent experiments.

Next, the effect of accumulation time was optimized (Fig. 7a). This pre-concentration step allows the high-surface-area, mesoporous electrode to adsorb ZEA from the solution prior to the DPV scan, thereby amplifying the signal. The plot of peak current versus time (Fig. 7b) shows that the DPV signal increases rapidly as more ZEA accumulates on the electrode surface, and then begins to plateau after approximately 180 s as the active sites approach saturation. An accumulation time of 180 s was thus chosen as the optimal balance between high sensitivity and rapid analysis time.

The analytical performance of the optimized Bi₂S₃-CF hybrid was benchmarked against the control electrodes (Fig. 8). In the presence of 100 ng/mL ZEA, the Bare CF electrode shows no discernible oxidation peak, demonstrating that it has no catalytic activity towards ZEA [22]. The drop-cast Bi₂S₃/CF control exhibits only a very weak and broad oxidation peak, with a current of ∼1.5 μA. This poor response is a direct result of its low ECSA, high R_ct, and non-uniform surface. In stark contrast, the Bi₂S₃-CF hybrid produces a strong, sharp, and well-defined oxidation peak at ∼+0.82 V, with a peak current of ∼14.8 μA. This represents a 10-fold signal amplification compared to the drop-cast control. This figure is the ultimate “payoff” of the study: it proves that both components are required in a synergistic configuration. The superior performance of the hybrid is a direct and quantifiable consequence of the high ECSA, low R_ct, and mesoporous structure created by the in-situ fabrication method [31].

The sensitivity of the Bi₂S₃-CF sensor was then evaluated by recording DPV responses for increasing concentrations of ZEA under optimal conditions (Fig. 9a). The oxidation peak current at +0.82 V clearly increases in a dose-dependent manner. Fig. 9b shows the resulting calibration plot, where the peak current is plotted against the logarithm of the ZEA concentration. The plot is highly linear, with a linear regression equation of I_p (μA) = 1.8 + 5.5 × log[ZEA (ng/mL)] and an excellent correlation coefficient (R²) of 0.999. The linear dynamic range was determined to be from 0.01 ng/mL to 200 ng/mL. The limit of detection (LOD) was calculated using the 3σ/slope approach, where σ represents the standard deviation of the blank response obtained from repeated DPV measurements in 0.1 M PBS (pH 6.0) containing 0 ng/mL ZEA under the same accumulation and scan conditions. The blank trace shows only a non-faradaic background current without a discernible oxidation peak at ~+0.82 V; therefore, σ was computed from the baseline-corrected current at the ZEA peak potential (i.e., the analytical readout used for calibration). The LOD was found to be 0.003 ng/mL (3 pg/mL) [22]. This ultralow detection limit is several orders of magnitude lower than the stringent regulatory limits set by the EFSA (e.g., 50 μg/kg = 50 ng/mL), confirming the sensor is exceptionally well-suited for food safety screening.

To contextualize this performance, the Bi₂S₃-CF sensor was compared with other electrochemical sensors for ZEA reported in the literature (Table 2). Our sensor demonstrates a linear range and LOD that is highly competitive with, or superior to, other state-of-the-art sensors, many of which require complex fabrication steps or biological recognition elements like aptamers [31]. It vastly outperforms simple, unmodified electrodes [32]. This comparison validates our in-situ hybrid material as a top-tier sensing platform.

A practical sensor must not only be sensitive but also selective, reproducible, and stable. The selectivity of the Bi₂S₃-CF sensor was evaluated by measuring its DPV response to 100 ng/mL ZEA in the presence of a 100-fold excess (10,000 ng/mL) of potential interfering species commonly found in food. These included other mycotoxins (Aflatoxin B1, Ochratoxin A, Fumonisin B1) and other electroactive molecules (ascorbic acid, glucose) [37]. The quantitative data in Table 3 show that none of these interferents produced a significant signal, with all relative responses being less than 5% of the ZEA signal. This demonstrates the high selectivity of the sensor, which is likely due to a specific electrocatalytic affinity of the Bi₂S₃ nanorod surface for the ZEA molecule’s phenolic structure.

The reproducibility of the sensor was assessed through intra-assay and inter-assay precision tests. The intra-assay relative standard deviation (RSD) for 10 consecutive measurements on a single electrode was found to be 1.94% (Fig. 10a). The inter-assay RSD, determined by measuring the same ZEA concentration on ten independently fabricated Bi₂S₃-CF electrodes (Fig. 10b), was 3.1% [38]. These low RSD values (<5%) confirm that the in-situ hydrothermal fabrication process is highly reliable and produces electrodes with consistent performance. The long-term stability of the sensor was monitored over 30 days, with the electrode stored at 4°C when not in use. As shown in Fig. 10c, the sensor retained over 95% of its initial DPV signal after 30 days. This excellent stability is attributed to the robust, chemically-bonded nature of the in-situ hybrid, which prevents the catalytic Bi₂S₃ material from delaminating or leaching from the CF substrate [38].

Finally, the practical applicability of the sensor was validated in real cereal matrices using a standard addition quantification protocol, which is widely adopted to minimize matrix effects in food analysis. In each matrix extract, ZEA was quantified from the standard-addition regression (x-intercept approach), and the same procedure was applied to spiked extracts to calculate recoveries (Table 4). A representative standard addition plot obtained in corn flour extract is shown in Fig. 10d, demonstrating linear response in the matrix and enabling accurate back-calculation of ZEA concentration. ZEA-free extracts from corn flour, wheat flour, and oat flakes were spiked with known concentrations of ZEA. Fig. 10d shows a representative standard addition plot for a corn flour extract, where the linear increase in DPV signal allows for accurate quantification of the spiked amount. The complete recovery data is summarized in Table 4. The recovery values for all samples ranged from 97.2% to 103.4%, with RSDs below 3.2% [5]. These excellent results demonstrate that the complex matrix of real cereal extracts does not interfere with the sensor’s accuracy. This validation confirms that the Bi₂S₃-CF hybrid electrode is a robust and reliable tool for the practical, on-site screening of ZEA in food.