The chemicals utilized in this work were all of analytical purity and required no additional treatment. Specifically, bovine serum albumin (BSA), cetyltrimethyl ammonium bromide (CTAB, 99%), vitamin B₂, and vitamin B₆ were procured from Maclin Biochemical Technology Co., LTD (Shanghai, China). Sodium hydrogen phosphate heptahydrate (Na₂HPO₄•7H₂O), sodium dihydrogen phosphate (NaH₂PO₄), tetraethyl orthosilicate (TEOS, 98%), potassium hydrogen phthalate (KHP), potassium ferricyanide (K₃Fe(CN)₆), sodium hydroxide (NaOH), hexammine ruthenium (III) trichloride (Ru(NH₃)₆Cl₃), starch, glucose, ascorbic acid (AA), dopamine (DA), and uric acid (UA) were acquired from Aladin Reagent Co., LTD (Shanghai, China). Ethanol, sodium nitrate (NaNO₃), sodium chloride (NaCl), calcium chloride (CaCl₂), potassium chloride (KCl), ferric chloride (FeCl₂), and magnesium chloride (MgCl₂, 95%) were sourced from Gaojing Fine Chemical Co., LTD (Hangzhou, China). Concentrated hydrochloric acid was obtained from Shuanglin Chemical Reagent Co., LTD. The ultrapure water (18.0 MΩ•cm) used in the experiments was prepared by Mill-Q Systems (Millipore). Multivitamin B tablets were acquired from the local pharmacy store (Hangzhou, China). Milk samples were obtained from the local supermarket (Hangzhou, China).

The morphology and thickness of VMSF were characterized using an HT7700 transmission electron microscope (TEM) from Hitachi, Japan. The sample preparation involved careful scraping off the VMSF layer from p-GCE using a knife and ultrasonical dispersing it in anhydrous ethanol. The resulting supernatant drops were coated onto a copper grid, dried, and then observed in the TEM chamber at an accelerated voltage of 200 kV. All electrochemical tests were conducted on an Autolab PGSTAT302N electrochemical workstation (Metrotron, Switzerland). The three-electrode system was employed, with the reference electrode being Ag/AgCl (saturated using KCl), the counter electrode being platinum wire or disk (2 cm × 4 cm), and the working electrode being either p-GCE or VMSF/p-GCE. The scan rate in the cyclic voltammetry (CV) test was set at 100 mV/s. For differential pulse voltammetry (DPV) measurements, the parameters including step potential (5 mV), pulse amplitude (25 mV), pulse time (0.05 s), and time interval (0.2 s) were set.

Initially, GCE underwent successive polishing with 0.5 μm, 0.3 μm, and 0.05 μm alumina powder. Following the polishing step, GCE surface was subjected to ultrasonic cleaning in ethanol and ultrapure water for 1 min each to remove residual alumina powder, ensuring a cleaner electrode surface. The newly polished GCE was activated using electrochemical polarization under static conditions. Briefly, a phosphate buffer solution (PBS, 0.1 M, pH = 5) served as the electrolyte. The electrochemical polarization involved both anodic oxidation and cathodic reduction (49). Anodic oxidation was carried out using a constant voltage of +1.8 V for 5 min. Subsequently, the obtained electrode underwent cathodic reduction at −1.0 V for 1 min, resulting in the electrochemically pre-activated GCE, referred to as p-GCE.

The modification of VMSF on p-GCE electrode was achieved through an electrochemical-assisted self-assembly (EASA) method (53). Initially, a precursor solution was prepared by weighing 1.585 g of CTAB and adding it to a mixed solution consisting of 20 mL ethanol and 20 mL NaNO₃. The mixture was stirred until completely dissolved. Subsequently, 3,050 μL of TEOS was added as the siloxane source, and the solution was stirred at room temperature for 2.5 h. For VMSF growth, the three-electrode system was immersed in the obtained precursor solution and a constant current (−0.7 mA/cm²) was applied on p-GCE for 10 s. Then, the resulting electrode was promptly removed and immersed in ultrapure water for thorough washing. After aging overnight at 80°C, the electrode with surfactant micelles (SM) blocking the nanochannels was obtained, denoted as SM@VMSF/p-GCE. Immersing it in a hydrochloric acid/ethanol solution to remove SM from the nanochannels resulted in an electrode modified with an array of open nanochannels (VMSF/p-GCE).

When detecting VB₂ with VMSF/p-GCE, a supporting electrolyte solution of PBS (0.1 M, pH = 7.0) was used. Different concentrations of VB₂ were introduced into the electrolyte solution, and the signals generated by the electrochemical oxidation reaction of VB₂ on the electrode were determined using the DPV method. For real sample analysis, VB₂ in milk or composite vitamin B tablets was determined. For milk samples, the samples were diluted 1,000 times with the supporting electrolyte solution before testing. For composite vitamin B tablets, the tablets were firstly ground into powder using a mortar and pestle, and 0.787 g of the powder was dispersed in ultrapure water (2 mL). Then, VB₂ content in the diluted milk or tablet leachate was determined using the standard addition method.

In this work, a simple electrochemical sensor was efficiently prepared for direct electrochemical detection of VB₂ by integrating VMSF with highly active p-GCE. As illustrated in Figure 1, the widely used GCE was chosen as the supporting electrode. However, GCE can not stably bind with VMSF and its electrochemical activity should be improved. Activating carbon-based electrodes through electrochemical polarization is a simple, green method for preparing highly active electrodes. This method requires no complex chemicals or cumbersome procedures and typically involves electrochemical pretreatment of the electrode in a conventional electrolyte solution. Thus, electrochemical preactivation of GCE through electrochemical polarization was performed to obtain p-GCE with high electrochemical activity and ability to stably bind with VMSF. Herein, electrochemical polarization was performed using anodic oxidation and cathodic reduction sequentially in a PBS solution. Specifically, a higher positive potential (+1.8 V) was applied to GCE for anodic polarization, followed by cathodic reduction at a negative potential (−1 V). During the anodic oxidation process, active oxygen radicals generated from water electrolysis oxidize and etch the sp²-conjugated carbon on GCE surface, leading to the formation of abundant edge carbon, defects, and oxygen-containing functional groups (54). X-ray photoelectron spectroscopy (XPS) characterization was employed to investigate the change of the surface chemical groups. The atomic percentages of carbon in GCE, the electrode after anodic polarization, and p-GCE are 14.6, 69.9, and 75.7%, respectively. The corresponding oxygen atomic percentages are 14.6, 30.0, and 24.3%. The changed atomic percentages indicates the changed chemical structure on the electrode surface. Supplementary Figure S1 displays high-resolution C1s spectra of GCE, the electrode after anodic polarization, and p-GCE. As shown, all electrodes have four carbon bonds including C-C/C=C(sp² C, 284.4 eV), C-O (285.8 eV), C=O (287.1 eV), and O-C=O (288.6 eV). After anodic polarization, the content of C=O and O-C=O bonds on electrode surface significantly increased, while the content of C-O bonds decreased, suggesting an increased oxidation degree of carbon atoms under high potential. Following the subsequent cathodic reduction, the content of C-O bonds increased, and the content of C=O bonds decreased significantly, indicating the reduction of C=O bonds to C-O bonds. Compared to GCE, the C-O bond content on p-GCE increased from 21.4 to 24.5%, indicating an increase in oxygen-containing groups on the GCE surface after electrochemical polarization. These functional groups serve as active sites, enhancing the adsorption of organic electroactive molecules on electrode and facilitating interfacial electron transfer reactions. Thus, compared to traditional GCE, p-GCE exhibits higher electrochemical and electrocatalytic activity.

The impact of electrochemical preactivation on electrode properties was examined using the standard redox probe Fe(CN)₆^3−/4− through cyclic voltammetry (CV). As shown in Figure 2A, Fe(CN)₆^3−/4− exhibits a distinct pair of oxidation/reduction peaks on GCE. After anodic polarization of GCE, there are almost no observable oxidation/reduction peaks on the obtained electrode, indicating that the high positive potential applied during anodic polarization imposes a strong oxidative effect on the electrode surface, inhibiting the electron transfer process at the electrode interface. Subsequently, after cathodic reduction of the electrode, oxidation–reduction peaks of Fe(CN)₆^3−/4− molecules reappear on p-GCE. Compared to GCE, p-GCE demonstrates higher charging current and oxidation–reduction peak current, indicating an increased electrode active area. Moreover, the peak-to-peak separation (66 mV) on p-GCE is smaller than that on GCE (75 mV), suggesting faster charge transfer kinetics for the redox probe on p-GCE.

To study the effect of electrochemical preactivation on potential resolution of electrode, the electrochemical characteristics of common electrochemical small molecules found in biological and food samples, including ascorbic acid (AA), uric acid (UA), and dopamine (DA) on GCE and p-GCE were further investigated. Figure 2B illustrates the differential pulse voltammetry (DPV) curves obtained on GCE and p-GCE in the analysis of the mixture of the three substances. As shown, GCE is unable to differentiate between the three electroactive components, detecting only a single mixed peak with a lower peak current. In contrast, p-GCE is able to distinguish between the three components, exhibiting higher peak currents and superior potential resolution capabilities. This may be attributed to the catalytic effect of oxygen-containing functional groups on the surface of p-GCE, thereby enhancing the electrode’s potential resolution capability.

GCE cannot directly and stably anchor VMSF. Processes such as washing after VMSF growth resulted in film detachment. Studies have shown that pre-treating with an organosilane adhesive layer can achieve stable VMSF modification (55). However, this operation increases the complexity of electrode preparation and, on the other hand, a non-conductive adhesive layer can reduce the electrode’s electrochemical activity. After preactivation of GCE, the successful integration of VMSF on p-GCE could be achieved without the use of any adhesives. This may be attributed to the formation of-O-Si-O-covalent bonds between the oxygen-containing functional groups on the surface of p-GCE and VMSF. As shown, the electrochemical assisted self-assembly (EASA) method was employed on p-GCE to further integrate VMSF. Currently, the most common methods for VMSF growth are the Stöber solution growth method and EASA method. The former can achieve large-area VMSF preparation on indium tin oxide (ITO) electrode but requires a longer preparation time (over 12 h). On the other hand, EASA method allows for rapid VMSF preparation on various electrodes, typically within 10 s. Based on the size exclusion effect and enrichment of VB₂ of VMSF nanochannels, direct and highly sensitive electrochemical detection of VB₂ in complex samples could be realized.

By examining the electrochemical performance of two standard redox probes with opposite charges on different electrodes, the integrity and charge-selective permeability of VMSF were investigated. Figures 3A,B show the CV curves obtained on p-GCE, SM@VMSF/p-GCE, and VMSF/p-GCE electrodes in solutions containing the anionic probe Fe(CN)₆³⁻ or the cationic probe Ru(NH₃)₆³⁺, respectively. Both probes exhibit well-defined oxidation/reduction peaks on p-GCE. When SM@VMSF/p-GCE is used as the working electrode, almost no Faraday current is detected on the electrode. This is attributed to the non-conductive VMSF covering the electrode surface on p-GCE, and the nanochannels of VMSF are filled with SM, keeping nanochannels in a closed state. As a result, redox probe cannot reach the electrode surface through the nanochannels, confirming the integrity of the VMSF film. In marked contrast, when VMSF is directly grown on GCE, the resulting SM@VMSF/GCE exhibits significant oxidation–reduction peaks for both probes (insets in Figures 3A,B). This is attributed to the weak binding between GCE and VMSF, causing the film to detach during cleaning processes after growth, thereby failing to completely cover the electrode surface. This also proves the importance of electrode preactivation to achieve stable binding of VMSF without the need for adhesives. Upon removing SM to obtain the VMSF/p-GCE electrode, oxidation/reduction peaks for both probes can be observed again. Compared to p-GCE, the peak current of the negatively charged Fe(CN)₆³⁻ probe on the VMSF/p-GCE electrode decreases, while the peak current of the positively charged Ru(NH₃)₆³⁺ probe significantly increases. This demonstrates the charge-selective permeability of the negatively charged surface of VMSF. The silica structure of VMSF is rich in silanol groups (pKa ~ 2), and their ionization in the buffer media results in a negatively charged surface. This leads to electrostatic repulsion toward negatively charged Fe(CN)₆³⁻ and significant electrostatic attraction for positively charged Ru(NH₃)₆³⁺.

When detecting complex food or drug samples with an electrochemical sensor, large molecules in the matrix often non-specifically adsorb to the electrode surface, thereby contaminating the electrode surface and affecting the sensitivity and reproducibility of detection. Using VB₂ as the electrochemical substance and bovine serum albumin (BSA) and starch as macromolecular models, the anti-fouling characteristic of p-GCE or VMSF/p-GCE electrodes were investigated. As shown in Figures 4A,B, when BSA or starch is present, the response of the p-GCE electrode to VB₂ is significantly reduced, indicating that these substances tend to contaminate the electrode and reduce the detection signal. In contrast, when these macromolecules are present, the DPV peak current of VB₂ obtained on the VMSF/p-GCE electrode remains almost unchanged, demonstrating the excellent anti-fouling capability. This anti-fouling ability is attributed to the size exclusion effect of the VMSF ultra-small nanochannels. Since these large molecules cannot enter the nanochannels, and the silica structure of VMSF is electrically insulating, BSA or starch can not contaminate the underlying electrode. This gives VMSF/p-GCE an advantage in the direct detection of complex samples.

The morphology and structure of VMSF were characterized using transmission electron microscopy (TEM). As shown in Figure 4C, the TEM image of VMSF in a top-down view clearly reveals hexagonally arranged mesoporous nanochannels. The film appears uniform, with no observable defects over a large observed area. Figure 4D displayed the TEM image of the cross-sectional section of VMSF with ordered nanochannels with a film thickness (length of nanochannel) of approximately 120 nm.

The electrochemical behavior of VB₂ on different electrodes was investigated using CV and DPV. ITO electrodes and GCE electrodes were employed as controls. Figures 5A,B depict the CV and DPV curves obtained on various electrodes in VB₂ solution. From the inset in Figure 5A, it can be observed that there are no apparent oxidation or reduction peaks for VB₂ on the ITO electrode. Only very little peak current signal is observed in the corresponding DPV curve (inset in Figure 5B). These results indicates that the ITO electrode is not suitable as a supporting electrode for detecting VB₂, an organic electroactive molecule. However, when GCE was used for detection, oxidation/reduction peaks of VB₂ appear in the CV, and an oxidation peak near-0.47 V is detected in the DPV curve. This is attributed to the presence of aromatic structure in the VB₂ structure, which can adsorb to the electrode surface through π-π stacking interactions, leading to an increase in peak current. For p-GCE obtained after electrode preactivation of GCE, both the oxidation–reduction peaks of VB₂ in CV curve and the oxidation peak current in DPV curve are significantly enhanced. This is because electrochemical preactivation results in a larger electroactive surface area of the electrode, with abundant edge carbon, defects, and oxygen-containing functional groups that can further promote the adsorption and electron transfer reactions of the analyte on the electrode surface through hydrogen bonding or electrostatic interactions. Upon modification of VMSF on the surface of p-GCE, the peak current of VB₂ measured on VMSF/p-GCE is further increased, demonstrating the significant enrichment effect of VMSF nanochannels on VB₂. Despite the enrichment effect of nanochannels, the signal of VB₂ on VMSF/ITO electrode is higher than that on ITO electrode but significantly lower than that obtained on VMSF/p-GCE. Thus, both the supporting electrode and the array of nanochannels on VMSF/p-GCE contribute to the enhanced electrochemical signal of VB₂.

The growth thickness of VMSF on the surface of p-GCE might affect detection sensitivity of VMSF/ p-GCE. As the thickness of VMSF is related to the deposition time in growth (26), the electrochemical signals of electrodes obtained under different VMSF deposition time were investigated in VB₂ solution. As shown in Supplementary Figure S2, the signal of VB₂ increases when VMSF deposition time increases from 2 s to 10s. Subsequently, with a further increase in VMSF deposition time, the electrochemical signal decreases. It is speculated that an excessively thick VMSF layer may compromise the detection performance in two aspects. Firstly, when the VMSF thickness is too thick, the mass transfer of the analyte from the solution to the electrode surface takes a longer time. Secondly, with an excessively long growth time of VMSF, silica nanoparticles might form on the outer surface of VMSF, partially blocking the nanochannels.

To maximize the sensitivity of VB₂ detection on VMSF/p-GCE, the detection conditions were optimized, including the pH of the supporting electrolyte and the VB₂ enrichment time. As shown in Figure 5C, with the increase in pH, the DPV oxidation current of VB₂ on the electrode gradually strengthened, reaching the maximum current response at pH = 7 (Figure 5C). Additionally, as the pH increased, the oxidation peak potential of VB₂ shifted toward negative potentials, indicating the involvement of protons in the electrochemical oxidation–reduction process of VB₂. From the inset in Figure 5C, it can be observed that the oxidation peak potential (E_pa) of VB₂ has a good linear relationship with pH (E_pa = −0.057 pH - 0.064, R² = 0.9991). The slope is close to that of the Nernst equation (59 mV/pH), suggesting that the electrochemical oxidation–reduction process involving VB₂ is an equivalent electron and proton transfer process. Thus, pH = 7 was chosen as the optimal detection condition. Under the optimal pH condition, time for VB₂ enrichment was investigated (Figure 5D). With the increase in enrichment time, the DPV oxidation current of VB₂ on the electrode gradually increased. After VB₂ was enriched for 180 s, the change of current gradually decreased. Considering both the peak current and the detection speed, VB₂ was enriched for 180 s for subsequent experiments.

Faradaic charge is a crucial indicator of electrochemical substances undergoing oxidation–reduction reactions on electrodes. To study the charge transfer of VB₂ on the sensing electrode, the relationship between the oxidation peak current and peak potential of VB₂ with the scan rate (ν) was investigated. As shown in Figure 6A, with the increase in scan rate, the oxidation–reduction peak currents of VB₂ gradually increased, and the peak potential shifted. The oxidation peak current (I_pa) and reduction peak current (I_pc) exhibited a good linear relationship with the scan rate (inset in Figure 6A, I_pa = 0.309ν − 1.77, R² = 0.9999; Ipc = −0.307ν + 2.86, R² = 0.9988). This indicates that the electrochemical reaction process of VB₂ molecules on the VMSF/p-GCE electrode is adsorption-controlled. The potential of oxidation peak (E_pa) exhibits a well-defined linear relationship with the logarithm of the scanning rate (logν) (Figure 6B), fitting to the following linear equation:

Here, E₀’ is the standard electrode potential, k_s is the reaction standard rate constant, D is the diffusion coefficient of VB₂, α is the electron transfer coefficient of VB₂, and n_α is the electron transfer number. The calculated n_α is 2.28, approximately equal to 2, indicating that the reaction is a double electron transfer process. The redox reaction is illustrated in Figure 1.

Under optimal conditions, the performance of VMSF/p-GCE in electrochemical detection of VB₂ was investigated. Figure 6C illustrated the DPV curves obtained by VMSF/p-GCE in solutions with different concentrations of VB₂. It can be observed that as the concentration of VB₂ increases, the oxidation peak current gradually increases. As revealed in Figure 6D, the oxidation peak current (I) and VB₂ concentration (C) exhibit a well-defined linear relationship in the range of VB₂ concentrations from 20 nM to 7 μM (I = 11.3C + 0.401, R² = 0.997), and from 7 μM to 20 μM (I = 0.797C + 70.8, R² = 0.996). The calculated detection limit (LOD) is 11 nM at a signal-to-noise ratio of 3 (S/N = 3). Table 1 outlines the detection performance of various electrochemical sensors for VB₂ (35–42). The LOD is lower than that obtained on nanoporous gold modified GCE (NPG/GCE) (35), reduced MoS₂-graphene/32-merhomoadenine ssDNA oligonucleotides modified Au electrode (rMoS₂-Gr/A₃₂/AuE) (37), magnetite nanoparticles/reduced graphene oxide modified GCE (Fe₃O₄NPs/rGO/GCE) (38), Co²⁺-Y zeolite modified carbon paste electrode (Co²⁺-Y/CPE) (40), Fe₃O₄NPs modified electrochemical paper-based analytical devices (Fe₃O₄NPs-ePADs) (41), carbon cloth supported two-dimensional MoS₂-MoO₃ modified screen-printed carbon electrode (2D-MoS₂-MoO₃-CC/SPCE) (42), but higher than that obtained on α-Fe₂O₃ nanoparticles/multi-walled carbon nanotube/gold nanoparticles modified GCE α-Fe₂O₃NPs/MWCNT/AuNPs/GCE (36), vitamin B₂-poly-O-aminophenol/molecularly imprinted polymers modified GCE (VB₂-PoAP/MIPs/GCE) (39). The low LOD and high sensitivity, attributed to the dual enrichment effect of the supporting p-GCE and VMSF on VB₂.

When the electrode was used, the complex matrix in real samples might affect the analysis. For instance, large molecules in samples, such as protein, DNA, polysaccharide, can nonspecifically adsorb and occupy active sites on the electrode surface, leading to electrode fouling. This fouling of this electrode will result in changes in electrochemical signals, reducing the accuracy of electrochemical detection. In addition, the co-existed redox molecules might also affect the detection. The selectivity and anti-fouling property of VMSF/p-GCE in detecting VB₂ was investigated. Common metal ions (Na⁺, Ca²⁺, Mg²⁺, K⁺), large molecular, starch, and small molecules present in the human body including ascorbic acid-AA, uric acid-UA, vitamin B₆, dopamine-DA, glucose-Glu were selected as the possible interfering substances to examine their impact on VB₂ detection. As shown in Figure 7A, the addition of each above substance does not cause a significant change in the current signal of VB₂, demonstrating the excellent selectivity of VMSF/p-GCE in VB₂ detection. This is attributed to the distinct electrochemical potentials between the mentioned substances and VB₂ on p-GCE. The electrostatic repulsion toward negatively charged redox molecules including AA and UA, also improve the selectivity. Additionally, large molecule such as starch cannot enter the nanochannel or contact with the underlying electrode. Thus, VSMF-modified electrode demonstrates anti-fouling property. To investigate the repeatability of VB₂ detection among VMSF/p-GCE, five VMSF/p-GCE electrodes were prepared using different batches. The relative standard deviation (RSD) of the current signal for detecting VB₂ (5 μM) was 0.8%, demonstrating good reproducibility in electrode preparation (Supplementary Figure S3).

The regenerative capability of VMSF/p-GCE was also examined. Stirring the electrode in hydrochloric acid-ethanol solution (0.1 M) for 20 min can effectively remove residual VB₂ inside the channels, completing the electrode regeneration. After regeneration, the electrochemical signal measured in the supporting electrolyte was extremely low, indicating the effective removal of VB₂ from the channels during the regeneration process (Figure 7B). The electrode was reused five times for VB₂ detection after regeneration. Compared with the initial detection signal, the obtained peak currents are all higher than 98% with a relative standard deviation (RSD) of 0.5%, demonstrating excellent regenerative capability of the sensor.

The VMSF/p-GCE was stored at room temperature, and VB₂ detection followed by regeneration was performed daily. It was observed in that even after regeneration and storage for 5 days, the peak current measured by VMSF/p-GCE remained at 93.8% of the first-day measurement, proving the electrode’s high stability (Figure 7C).

To assess the accuracy and precision of the sensor in real sample detection, the standard addition method was employed to detect the VB₂ content in leachate of compound vitamin B tablet and milk samples. As shown in Table 2, the recovery rates ranged from 94.2 to 102%, with a RSD of less than 4.5% for three measurements, demonstrating the accuracy of the detection. Additionally, the VB₂ content in the leachate of compound vitamin B tablet was determined using the extrapolation method and found to be 0.318 μM (as shown in Figure 7D), which closely aligned with the value determined by the standard (HPLC GB5009852016). Due to its convenient preparation, simple operation, and the absence of complex sample pretreatment, the developed electrochemical sensor has significant potential for direct electroanalysis of VB₂ in complex samples.