Tetraethoxysilane (TEOS), cetyltrimethylammonium bromide (CTAB), potassium ferricyanide (K₃ [Fe(CN)₆]), potassium ferrocyanide (K₄ [Fe(CN)₆]), sodium phosphate monobasic dihydrate (NaH₂PO₄.2H₂O), sodium dodecyl sulfate (SDS), amylum, humic acid (HA), sodium phosphate dibasic dodecahydrate (Na₂HPO₄.12H₂O), and catechol (CC) were purchased from Aladdin. Bisphenol A (BPA), hydroquinone (HQ), bisphenol S (BPS), and 4-nitrophenol (4-NP) were purchased from Macklin. Hexaammineruthenium (III) chloride (Ru(NH₃)₆Cl₃) and bovine serum albumin (BSA) were purchased from Sigma Aldrich. Ethanol was obtained from Hangzhou Shuanglin Chemical reagent. Calcium chloride (CaCl₂), potassium chloride (KCl), sodium chloride (NaCl), magnesium sulfate (MgSO₄), and sodium nitrate (NaNO₃) were purchased from the Hangzhou Gaojing Fine Chemical Industry. Environmental water samples were obtained from the lake of Zhejiang Sci-Tech University (Hangzhou, China). Soil leaching solution (SLS) was obtained by leaching the soil (1 g from lawn of Zhejiang Sci-Tech University) in 100 ml ultrapure water. All chemicals and reagents were of analytical grade and used as received without further purification. Ultrapure water (18.2 MΩ cm) was used to prepare all aqueous solutions throughout this work.

Transmission electron microscopy (TEM) images were obtained at an acceleration voltage of 100 kV on a HT7700 transmission electron microscope (Hitachi, Japan). Before TEM measurement, the VMSF was gently scraped from the p-GCE surface and dispersed in ethanol by ultrasonication. Then, VMSF dispersion was dropped onto the copper grids. All electrochemical experiments including cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS), and differential pulse voltammetry (DPV) were conducted on an Autolab PGSTAT302N electrochemical workstation (Metrohm, Switzerland). A typical three-electrode system was adopted including bare or modified GCE as the working electrode, an Ag/AgCl electrode (saturated KCl) as the reference electrode, and a platinum wire electrode as the counter electrode. The scan rate in CV is 50 mV/s, unless particularly indicated. For DPV measurements, the step, modulation amplitude, modulation time, and interval time were 0.005 V, 0.025 V, 0.05 s, and 0.2 s, respectively.

The GCE (d = 3 mm) was first polished with 0.3 and 0.05 μm alumina power. The electrode was then sequentially cleaned by sonication in ethanol and ultrapure water and dried under nitrogen flow. Electrochemical activation includes anodic oxidation at high voltage and cathodic reduction at low voltage. Briefly, a constant potential (+1.8 V) was applied on the GCE for 300 s followed with a cyclic voltammetry scan (−1.3–1.25 V, scan segments: 6) in phosphate-buffered solution (PBS, 0.1 M, pH 5). Then, the obtained p-GCE was washed with ultrapure water and dried under nitrogen flow.

The VMSF was grown on p-GCE by using electrochemically assisted self-assembly (EASA) methods as previously reported. (Walcarius et al., 2007) Typically, NaNO₃ (20 ml, 0.1 M, pH = 2.6) and ethanol (20 ml) were first mixed (v:v = 1:1). Then, CTAB (1.585 g) and TEOS (2.833 g) were subsequently added under stirring. The aforementioned mixture was further stirred for 2.5 h to prehydrolyze TEOS to obtain the precursor solution. After p-GCE was immersed in the precursor solution, growth of the VMSF was performed by applying a cathodic current (−52.2 μA) to p-GCE for 10 s. Then, the obtained electrode was quickly removed from the precursor solution, followed by thorough rinsing with ultrapure water and dried under a N₂ stream. After further aging at 80°C overnight, the modified electrode containing surfactant micelles (SMs) inside the nanochannels was obtained and termed as SM@VMSF/p-GCE. The removal of SMs could be realized by treating SM@VMSF/p-GCE with a 0.1 M HCl-ethanol solution under stirring for 5 min. The resulting electrode with open nanochannels was termed as VMSF/p-GCE.

Phosphate buffer solution (PBS) (0.1 M, pH = 6) was applied as the buffer for the detection of BPA. The electrochemical responses of different concentration of BPA were recorded using CV or DPV. For real sample analysis, environmental water samples were diluted using the buffer by a factor of 10 without other pretreatments, and the soil was dispersed in the buffer to form a suspension of 1 mg/ml, whose supernatant was further adopted. Then, BPA detection in complex real samples was achieved by the VMSF/p-GCE platform.

As illustrated in Figure 1, the electrochemically activated GCE (p-GCE) is applied as the supporting electrode to grow the vertically ordered mesoporous silica-nanochannel film (VMSF). Compared with the GCE, p-GCE has been proven to possess abundant active sites including edge plane sites, defects, and oxygen-containing functional groups. (Li et al., 2019b) These sites can not only enhance the adsorption of electroactive organic molecules (e.g. through electrostatic interaction, hydrogen bonding, etc) but also promote the electron transfer reaction, demonstrating attractive electrocatalytic activity. (Shi and Shiu, 2002) In addition, oxygenated groups on the surface of the p-GCE electrode (e.g. hydroxyl groups) can react with the silanol group so that the VMSF can stably bind on the electrode surface. The VMSF was rapidly and stably grown on p-GCE by the electrochemically assisted self-assembly (EASA) method. The EASA method is a convenient strategy for the fast preparation of the VMSF within 10 s. The principle is to apply a negative voltage on the electrode to promote the reduction of water to generate hydroxyl ions. Then, the pH gradient generated on the surface of the electrode promotes the self-assembly and condensation reactions of siloxanes around surfactant micelles (SMs). After VMSF growth, SMs block nanochannels (SM@VMSF/p-GCE). After further removal of SMs, an electrode with open nanochannels can be obtained (VMSF/p-GCE).

The CV curves of GCE, p-GCE, and VMSF/p-GCE recorded in pure PBS (0.1 M, pH = 6) without any electroactive species are shown in Supplementary Figure S1A (in supporting information-SI). Supplementary Figure S1B (SI) gives the relationship between the peak current obtained on GCE vs. the square root of scan rate. The non-Faradaic double layer capacitance (C _dl) can act as quantitative indicator of the surface area that is accessible to the electrolyte ions, which can indicate the electrochemical active surface area (ECSA) of electrodes. (Wei et al., 2019) A notably increased capacitive current is observed on p-GCE compared with that of the GCE (about ∼4 fold increasing), suggesting an enlarged ECSA owing to the generation of a thick porous layer during the electrochemical activation process. (Shi and Shiu, 2002) Interestingly, the ECSA of VMSF/p-GCE only decreases very slightly after the covering of the VMSF. On the other hand, a pair of redox peaks near ∼0 V is observed on p-GCE and VMSF/p-GCE, which are resulted from the conversion between surface-bonded quinone and hydroquinone generated during the electrochemical polarization pretreatment. (Engstrom and Strasser, 1984) Furthermore, p-GCE and VMSF/p-GCE demonstrate an improved electroanalytical reactivity proven by the larger decomposition currents and reduced decomposition potentials for both the anodic and cathodic limits. The exact ECSA of the GCE can be calculated using a reversible probe K₃ [Fe(CN)₆] by Randles–Sevcik equation. (Alam and Deen, 2020) The ECSA of the GCE is calculated to be 0.0584 cm². For comparison, the ECSA of p-GCE and VMSF/p-GCE are 0.222 and 0.204 cm², respectively. Thus, the electrochemical polarization increases the active area, and the equipment of the VMSF on p-GCE does not significantly decrease the active area of the electrode.

The integrity and permeability of the VMSF were investigated by electrochemical methods. The electrochemical signals of the standard redox probe (Ru(NH₃)₆ ³⁺) on different electrodes including the GCE, p-GCE, VMSF/p-GCE, and SM@VMSF/p-GCE are shown in Figure 2A. The anodic peak potential (E _pa), cathodic peak potential (E _pc), peak-to-peak separation (ΔE), anodic peak current (I _pa), and cathodic peak current (I _pc) of all the electrodes are demonstrated in Supplementary Table S1 (SI). As shown, GCE shows a pair of redox peaks with an ΔE _p of 65.92 mV. For p-GCE, an increase in the peak current (I _pa, 6.813 μA; I _pc, −6.022 μA) and a decrease in the ΔE _p (63.58 mV) are observed, indicating a faster electron transfer and enlarged ECSA. In the presence of hydrophobic CTAB micelles in the VMSF, there is no Faraday current response at SM@VMSF/p-GCE, indicating that the mass transfer of the hydrophilic probe to the electrode surface is inhibited. The failed electron transfer between the probe and the electrode proves that the VMSF grown on p-GCE is intact with no defects. When the SM is removed and p-GCE is modified with open nanochannels, an increased ΔE _p of VMSF/p-GCE (65.92 mV) is observed compared with that of p-GCE. However, VMSF/p-GCE demonstrates an enhanced peak current (I _pa, 10.18 μA; I _pc, −9.449 μA). This can be attributed to the strong electrostatic interaction between negatively charged silanols (pK _a of ∼2) on the nanochannel surface and positively charged Ru(NH₃)₆ ³⁺ at high pH and low ionic strength conditions. (Yan et al., 2020).

Consistent results were obtained from electrochemical impedance spectroscopy (EIS) in the presence of a standard anionic probe (Fe(CN)₆ ^3/4−). The Nyquist plots of GCE, p-GCE, and VMSF/p-GCE were measured under open-circuit voltage to further study the electrochemical behaviors of different electrodes (Figure 2B). The insets are the schematic illustration of the equivalent circuit (left inset) and the enlarged view of curves in a high-frequency region (right inset). The equivalent circuit contains the solution resistance (R _s), double layer capacitance (C _dl), Warburg impedance (Z _w), and apparent charge transfer resistance (R _ct). As shown, each electrode exhibits a semicircle in the high-frequency region representing the electron transfer–limiting process and a linear portion at the low-frequency region corresponding to the diffusion-limited process. As known, the effective diameter of the semicircle in a high-frequency region is equal to R _ct which is responsible for electron transfer kinetics of redox reactions at the electrode–electrolyte interface. Supplementary Table S2 displays the R _s and R _ct obtained on different electrodes. As shown, p-GCE shows the lowest R _ct (61 Ω), suggesting a faster charge transfer kinetics at the electrode interface after electrochemical pretreatment (the R _ct of GCE is 82 Ω). After p-GCE is covered by SM-blocked nanochannels, a remarkably increased R _ct (407 Ω) is observed on SM@VMSF/p-GCE owing to the inhibited probe diffusion. When the SM is removed, the R _ct (302 Ω) of VMSF/p-GCE further decreases owing to the permeability of the high-density nanochannels of the VMSF. However, it is still higher than that of p-GCE, which is due to the electrostatic repulsion between negatively charged VMSF⁻ and Fe(CN)₆ ^3-/4.

As demonstrated by transmission electron microscopy (TEM) images, the VMSF possesses homogenously distributed nanopores with a hexagonally packed structure and uniform diameter between 2 and 3 nm (Figure 2C). The porosity is ∼45%. The cross-sectional TEM image proves the perpendicular orientation of nanochannels to the substrate (Figure 2D).

The electrochemical behavior of BPA at the VMSF/p-GCE, p-GCE and GCE was investigated. As shown in Figure 3A, BPA shows an irreversible oxidation process at these three types of electrodes, which is consistent with many previous reports. (Pang et al., 2020; Freitas et al., 2020; Hu et al., 2016) In comparison with the GCE, p-GCE exhibits a rather higher peak current. This is mainly ascribed to the increase of the electroactive surface and electrocatalytic ability through electrochemical activation (Figure 3B). (Xuan et al., 2021) Owing to the interaction between the abundant silanol groups on the nanochannel surface and phenolic hydroxyl groups on BPA through hydrogen bond, the oxidation current of BPA further increases after equipment of the VMSF on p-GCE (Figure 3B). Due to the enrichment effect from both the supporting electrode and nanochannels, the VMSF/p-GCE can realize dual-signal amplification, leading to high sensitivity in detection.

To achieve an excellent performance toward the analysis of BPA, the detection conditions were optimized. As shown in Figure 3C, the peak current first increases with the increase of pH and reaches the highest value at pH 6. Then, the peak current decreases with the increase of pH. The decrease of the peak current at higher pH values might be attributed to the electrostatic repulsion of the negatively charged VMSF toward anionic BPA (pK _a1 = 9.6, pK _a2 = 10.2). The decline of peak current in an acid medium might result from the present H⁺ ion, which is one of the products of electro-oxidation of BPA, posing an inhibiting effect on the oxidation peak current. In addition, E _pa shifts negatively in the investigated pH range (Figure 3D). A good linear relationship is revealed between E _pa and the pH value (E _pa = −0.055 pH + 0.9919, R ² = 0.9980). The number of protons and electrons involved in the oxidation of BPA is calculated using the following equation (Eq. 1): d E p a d p H = 2.303 m R T n F , (1) where R is the gas constant (R = 8.314 J mol⁻¹ K⁻¹), T is the absolute temperature (T = 298 K), F is the Faraday constant (F = 96,485 C mol⁻¹), and m and n are the number of protons and electrons, respectively. The ratio of m/n is calculated to be 0.93, indicating that the number of protons and electrons involved in the oxidation of BPA is same. To further study the mechanism of the electro-oxidation process of BPA at VMSF/p-GCE, the electrochemical signal of BPA at different scan rates is investigated (Figure 3E). A linear relationship between peak current (I) and scan rate (v) is found (I = 0.0799v + 0.0855, R ² = 0.9899), indicating that the electro-oxidation of BPA at the VMSF/p-GCE is adsorption-controlled. The relationship between peak potential (E _pa) and scan rate (v) in a completely irreversible process can be described according to the following Laviron equation (Eq. 2): E p a = E o + ( R T α n F ) ln R T K s α n F + ( R T α n F ) ln ⁡ v , (2) where E ^o is the formal redox potential; α is the transfer coefficient, which is assumed to be 0.5 in a total irreversible electrochemical reaction; n is the number of electron transfers involved in the rate-determining step; and k _s is the standard rate constant of the reaction. The other quantities are the same as mentioned earlier. Accordingly, the relationship between E _pa and v can be depicted as E _pa = 0.0261l n v + 0.0546 (R ² = 0.9905, Figure 3F). The value of n is calculated as 1.97. Therefore, the oxidation of BPA on the VMSF/p-GCE is a 2H⁺/2e⁻ transfer process, which can be described as the following equation (Eq. 3). 3

Since the oxidation of BPA on VMSF/p-GCE is adsorption-controlled as mentioned previously, the accumulation time before detection is further optimized. As shown in Figure 4A, the oxidation peak current increases when the stirring time is increased and then reaches a platform at 100 s. Thus, an accumulation time of 100 s is chosen for further investigation.

Figure 4B shows the differential pulse voltammetry (DPV) curves obtained on the VMSF/p-GCE in the presence of different concentrations of BPA. A good linear correlation was found between the peak current (I) and the concentration of BPA (C _BPA) from 50 nM to 1.0 μM (I = 1.348C _BPA − 0.0425, R ² = 0.9915) and 1.0–10.0 μM (I = 0.7571C _BPA + 0.7130, R ² = 0.9938). The limit of detection (LOD) is calculated to be as low as 15 nM, at a signal-to-noise ratio of 3(S/N = 3). The comparison between the determination of BPA using different electrodes is demonstrated in Table 1. The LOD is lower than that obtained from graphene oxide-poly (1-[3-(N-pyrrolyl) propyl]-3-butylimidazolium bromide)–modified GCE (GO-poly (NPBimBr)/GCE) (Wang et al., 2018), tyrosinase-graphdiyne-chitosan–modified GCE (Tyr-GDY-CS/GCE) (Wu et al., 2020), Ag nanoparticle/multiwalled carbon nanotube–modified GCE (AgNP/MWCNT/GCE) (Goulart et al., 2018), reduced graphene oxide-Fe₃O₄/chitosan/laccase–modified GCE (rGO-Fe₃O₄/CS/laccase/GCE) (Fernandes et al., 2020), and nanoporous PtFe/graphene–modified GCE (NP-PtFe/Gr/GCE) (Tian et al., 2018) but higher than that obtained on Cu₂O-CuO@graphene quantum dot–modified GCE (Cu₂O-CuO@GQD/GCE) (Ashraf et al., 2020), GO-MWCNT-β-cyclodextrin–modified screen-printed carbon electrode (GO-MWCNT-βCD/SPE) (Alam and Deen, 2020), molecularly imprinted polymer/polypyrole-modified laser-scribed graphene (MIP/PPy@LSG) (Beduk et al., 2020), and NiS/rGO-modified mechanical pencil lead (NiS/rGO/MPL). (Vu et al., 2019).

The detection selectivity is critical for the real application of the electrochemical sensor. To evaluate the selectivity, the performance of the developed VMSF/p-GCE sensor to detect BPA in the presence of common metal ions (K⁺, Na⁺, Ca²⁺, and Mg²⁺) and electroactive environmental pollutants (BPS, CC, HQ, and 4-NP) is investigated. As shown in Figure 4D, these co-existed molecules do not interfere with the detection of BPA. The high selectivity is attributed to the good potential resolution ability of p-GCE. On the one hand, the co-existed ions cannot be oxidated in the applied potential range. On the other hand, electroactive organic molecules have different electrochemical behaviors and different electrocatalytic oxidation potentials on p-GCE owing to its good electrocatalytic activity. Thus, the detection of BPA has good selectivity. In addition, polysaccharides (starch), proteins (BSA), organic macromolecules (humic acid, HA), and surfactants (sodium dodecyl sulfate, SDS) that usually exist in complex environmental samples are selected as model substances to evaluate the anti-interference ability of the VMSF/p-GCE sensor in complex matrices. Figure 5 shows the ratio of current responses (I/I ₀) to BPA before (I ₀) and after (I) incubation of p-GCE or the VMSF/p-GCE with one of the possible interferences for 10 min. As seen, the current signals of p-GCE significantly decrease by nearly 50% in the presence of the interferences, indicating a severe matrix effect. Thus, the substances that are frequently present in complex samples can remarkably passivate the electrode, leading to a serious change of sensitivity. In contrast, almost no significant current change is observed for the VMSF/p-GCE, suggesting an excellent anti-smudge ability. This is attributed to the fact that the nanochannels of the VMSF can protect the electrodes from severe contamination in complex matrices through size repulsion and charge repulsion effects, which endow the VMSF-modified electrodes with a good antifouling ability. Thus, the VMSF/p-GCE sensor has great potential for the direct detection of complex samples. In comparison with other optical or chromatographic strategies (e.g. analysis), electroanalysis has advantages of simple instruments, fast detection, and high selectivity. (Cui et al., 2020; Zhao et al., 2020).

The repeatability, reproducibility, and stability of the developed VMSF/p-GCE sensor are also examined (Figures 6A–C). The repeatability was evaluated by detecting BPA (1.0 μM) five times using the same electrode. The electrode is easily regenerated by immersing in an HCl–ethanol (0.1 M) solution for 5 min. A relative standard deviation (RSD) of 1.4% is found, suggesting satisfactory repeatability (Figure 6A). To investigate the reproducibility of the sensor, five electrodes are parallelly fabricated under the same conditions. A low RSD of 0.7% is revealed for detecting BPA (1.0 μM), indicating high reproducibility (Figure 6B). The stability of the developed sensor is investigated by comparing the detection of BPA (1.0 μM) before and after storage in 4°C for 7 days. As shown in Figure 6C, the peak current is 95.0% that of the freshly fabricated electrode, confirming high stability of the fabricated sensor.

Considering the excellent antifouling ability of the VMSF/p-GCE sensor, the direct analysis of environmental water and soil leaching solutions is investigated using a standard addition method. As shown in Table 2, the detection exhibits satisfactory recoveries ranged from 98.0–104.5%, with a low relative standard deviation (RSD≤ 3.9%), indicating high reliability. As demonstrated in the inset of Figure 6D, the analyzed soil leaching solution is a suspension. However, the detected concentration of artificially added BPA (0.99 μM) by the extrapolation method (I = 0.810C + 0.806, R ² = 0.9978) is quite similar with the added concentration (1.0 μM). These prove the reliability of the VMSF/p-GCE for direct BPA analysis in real complex samples without separation.