Antifouling bismuth composite for robust electrochemical detection of heavy metals in complex matrices

Abstract

Electrochemical detection of heavy metals in complex matrices, such as biofluids, food, and water, using bismuth-based composites allows for point-of-care diagnostics in healthcare, food safety, and environmental monitoring. However, challenges remain in commercialization, particularly due to electrode fouling and sensitivity loss. This study introduces a robust antifouling coating consisting of a 3D porous cross-linked bovine serum albumin (BSA) matrix and 2D g-C₃N₄, supported by conductive bismuth tungstate. This composite effectively prevents nonspecific interactions, enhances electron transfer, and maintains 90% of the signal after one month in untreated human plasma, serum, and wastewater. It enables sensitive and multiplexed detection of heavy metals in plasma, serum, and water, providing a simple, stable, and efficient solution for the development of electrochemical sensors.

Introduction

Electrochemical sensors based on deposition-stripping analysis are used for economical, highly sensitive, and high-throughput detection of heavy metals, such as the widely commercialized dropping mercury electrode. However, due to the toxicity of mercury, many new analytical materials have been used as replacements, such as bismuth metal^1,2,3,4 and bismuth alloys^5,6, which have become prominent in this field. These sensors are particularly valuable for multi-point detection in complex matrices like food, blood, and tissue fluids⁷. To enhance heavy metal detection performance (sensitivity and selectivity), three key strategies have emerged: (1) increasing chelation of target ions on the electrode surface⁸; (2) improving fixation efficiency of reduced heavy metal atoms via alloy formation⁹; and (3) mitigating interference from environmental substances (e.g., organic compounds)¹⁰. Consequently, research has focused on elucidating chelation mechanisms, optimizing electrodeposited metal fixation, and enhancing electrode antifouling properties¹¹.

Functional material modification is critical for boosting ion chelation and capture¹². Conductive materials doped with elements (e.g., sulfur¹³ and nitrogen¹⁴) bearing lone electron pairs are effective: their Coulombic interactions with electro-deficient heavy metals enhance chelation¹⁵, enriching ions at the electrode surface for pre-deposition¹⁶. Additionally, such materials improve adhesion of electrodeposited metals¹⁷, strengthening fixation and overall detection performance¹⁸.

Another effective strategy is to use functional materials that can form alloys with heavy metal elements after electrodeposition and reduction^19,20,21,22. Bismuth metal exhibits a wide potential window, low background current, and alloy-forming ability comparable to mercury²³. However, bismuth films are prone to hydrolysis under alkaline conditions, limiting practical use. This has driven the adoption of bismuth compounds (such as Bi₂O₃²⁴, Bi₂WO₆²⁵, BiVO₄²⁶, BiFeO₃²⁷, Bi₂CuO₄²⁸, etc.), whose stable crystal structures balance electrochemical activity and reusability.

The successful use of bismuth-based surface-modified electrochemical sensors for heavy metal detection has been reported in numerous scientific studies²⁹. However, the commercialization of this technology has been limited due to its inability to maintain stable sensing performance in complex media environments. Unlike optical sensing, the challenge in electrochemical heavy metal sensing lies in the electrode configuration and the closed electrical and ionic circuits formed between the instrument and the sample, which are coupled to the interrogation and reading process. As a result, any material that nonspecifically binds to the electrode in a complex sample will reduce both the current and sensitivity. To address the challenge of high-sensitivity, high-throughput detection and analysis of multiple heavy metals in complex samples, we developed a three-dimensional (3D) nanocomposite. This structure integrates BSA crosslinked with conductive two-dimensional nanomaterials to form ion transport channels for heavy metals³⁰. The selected conductive material, 2D g-C₃N₄ (Fig. 1b), enhances electron transfer to the electrode, improves analytical performance, reduces nonspecific binding, and facilitates the efficient capture of heavy metal ions. We encapsulated the bismuth-based composite within this 3D structure to enhance the fixation and complexation of the target metal ions after electroreduction deposition.

Results

Design and preparation of conductive polymer film electrodes

The preliminary experiments began with the study of conductive cross-linked membranes with antifouling properties (Fig. 1a). Bovine serum albumin and g-C₃N₄ were used as the main functional monomers, glutaraldehyde served as the cross-linker for polymerization, and flower-like bismuth tungstate was added as a heavy metal co-deposition anchor to the pre-polymerization solution. The above pre-polymerized solution is uniformly dispersed by mixing and ultrasonic treatment, and then immediately dropped onto the electrode surface to form a coating. The performance of the constructed electrochemical sensor was evaluated using cyclic voltammetry, and the electrode was cyclically scanned between oxidation and reduction potentials in a standard potassium ferrocyanide-potassium ferrocyanide redox system. The potential difference (ΔE_p) and the corresponding current density of the redox process peak were analyzed to evaluate the electron transfer kinetics at the electrode-solution interface, thereby gaining a deeper understanding of the state of the electrode solid–liquid interface.

Fig. 1: Ion-scale transport channel electrochemical analysis platform.

a Schematic diagram of a constructed three-electrode electrochemical device, wherein the working electrode of gold is functionally modified with a conductive polymer membrane formed by BSA/Bi₂WO₆/g-C₃N₄/GA, with ion transport channels within the polymeric membrane. b Typical cyclic voltammograms showing oxidation and reduction curves of 5 mM ferroc/ferricyanide equimolar solutions for coated electrodes modified with various complexes (left) as well as polymeric film coated electrodes after crosslinking with 5% GA (right). c The electrochemical performance of electrodes with different coatings was evaluated. The bar chart illustrates the average current density of different nanocomposite-coated electrodes before (orange) and after (green) being exposed to 10 mg/mL HSA for 1 day. The line graph represents the final average peak-to-peak distance, based on measurements from five independent electrodes (n = 5). d Cyclic voltammograms of the gold working electrode without any modification (left) and the BSA/Bi₂WO₆/g-C₃N₄/GA coated electrode (right) in 5 mM equimolar ferric/ferrocyanide solution, with the scan rate set in the range of 0.01–0.1 V s⁻¹. e Oxidation/reduction peak currents (i_p) were extracted from the cyclic voltammograms in (d), and the values obtained from five independent electrode experiments were averaged (blocks). These values were then plotted against the square root of the scan rate. The error bars in all figures represent the standard deviation of the mean.

While BSA coating resulted in complete passivation, BSA/Bi₂WO₆, BSA/g-C₃N₄, BSA/NH₂-rGO, BSA/Bi₂WO₆/g-C₃N₄, and BSA/Bi₂WO₆/NH₂-rGO coatings retained 42%, 53%, 49%, 75%, and 68% of the current density, respectively (Fig. 1b, c), partly due to the electron transfer mediated by Bi₂WO₆, NH₂-rGO, and g-C₃N₄ particles. To evaluate the antifouling properties of these coatings, we measured their performance before and after 1 day of incubation in a 10 mg/mL human serum albumin (HSA) solution. The electrochemical performance of all coatings showed similar decreases, except for BSA/Bi₂WO₆/g-C₃N₄ and BSA/Bi₂WO₆/NH₂-rGO. The BSA/Bi₂WO₆/g-C₃N₄ composite maintained a high current and a low ΔE_p, while the BSA/ Bi₂WO₆ coating exhibited a much larger ΔE_p of up to 0.38 V. This indicates that fouling, particularly the blockage of Bi₂WO₆ pores, restricted the diffusion of iron/ferrocyanide to the electrode surface. The addition of NH₂-rGO and g-C₃N₄ two-dimensional layered materials can slightly alleviate the effect of electrode blockage by biomass, but still cannot eliminate the risk of fouling (Fig. 1c). To address this limitation, glutaraldehyde (GA) was introduced to crosslink BSA and g-C₃N₄ molecules, forming a 3D polymer matrix embedded with conductive nanomaterials. After cross-linking, the composite matrix of this polymer film significantly improved the electrochemical performance of the sensor, indicating that the cross-linking effect enhances the functionality and stability of the material. To explore the anti-fouling research of the electrode in clinical analysis, we considered that after incubating the prepared electrodes in 10 mg/mL HSA for 1 day, the BSA/NH₂-rGO/GA, BSA/ Bi₂WO₆/NH₂-rGO/GA, BSA/g-C₃N₄/GA and BSA/ Bi₂WO₆/g-C₃N₄/GA coatings retained 92% and 86%, 94% and 91% of the current density, respectively, and the ΔE_p was 218 mV, 229 mV, 128 mV and 190 mV, respectively. To evaluate the mass transfer of ferricyanide/potassium ferrocyanide on the prepared BSA/g-C₃N₄/GA and BSA/Bi₂WO₆/g-C₃N₄/GA coating electrodes, voltammograms were measured at different scan rates. The experimental results show that the collected voltammograms are almost consistent with those observed on the bare gold electrode (Fig. 1d). The final experimental results show that the redox current is linearly positively correlated with the square root of the scan rate, which indicates that the electrode surface presents a diffusion-limited process (Fig. 1e).

To gain a deeper insight into the mechanism behind the improved electrochemical performance of the polymerized film, the morphology of the coating and its polymerization mechanism were characterized using X-ray photoelectron spectroscopy (XPS) and scanning electron microscopy (SEM). The polymer film is formed by using BSA and g-C₃N₄ as polymerization monomers and GA as a cross-linking agent. The results show that when the GA content in the polymer precursor is low, powdery and dispersed dimers, oligomers, and even a large amount of polymer monomers with incomplete polymerization will be formed under low cross-linking conditions. These particles and molecules are adsorbed on the grooves and pores on the flower-like Bi₂WO₆ surface through physical interactions such as electrostatic forces (Fig. 2A). After cross-linking of the above-mentioned monomers with an appropriate ratio of GA, a thick porous sponge-like conductive polymer matrix was generated (Fig. 2B), which is similar to the aggregates observed in the previously reported BSA/GA protein antifouling matrix. The BSA/g-C₃N₄/Bi₂WO₆/GA coating, formed by introducing polymers to the surface of porous bismuth tungstate, exhibited the most significant microporosity. Notably, the simultaneous enhancement of antifouling performance and electrochemical activity stems from the synergistic effect of its porous structure and the incorporation of bismuth-based materials. Even with a film thickness exceeding 1 μm (Supplementary Fig. 1), the polymer coating retains efficient ion transport and electroreduction capabilities (Supplementary Fig. 2). These macroporosities were mainly contributed by the porous nature of the flower-like bismuth tungstate itself, in which the polymer with ion channels was tightly attached to the bismuth tungstate surface. Unfortunately, the pores of ion migration scale formed by the polymer could not be directly observed.

Fig. 2: Characterization of BSA/Bi₂WO₆/g-C₃N₄/GA nanocomposites with low (A) and appropriate (B) cross-linking degrees.

A Scanning electron micrograph at different scales and elemental mapping of BSA/Bi₂WO₆/g-C₃N₄/GA coated electrodes at low crosslinking degree between BSA and g-C₃N₄. Two sets of prepared samples were analyzed. B Scanning electron micrographs of BSA/Bi₂WO₆/g-C₃N₄/GA coated electrodes at appropriate cross-linking degrees at different scales and elemental mapping. Two sets of prepared samples were analyzed. C XPS graphs of BSA/Bi₂WO₆/g-C₃N₄/GA at different crosslinking degrees.

Considering the excellent electrochemical performances exhibited by BSA/NH₂-rGO/Bi₂WO₆/GA and BSA/g-C₃N₄/Bi₂WO₆/GA coated electrodes, we hypothesize that the prepared polymer coatings exhibit an extremely small electrode ensemble with ion size-restricted passage, and each ion channel acts as a working electrode for the ion migration scale. Similar behavior was also observed in the previously reported BSA-GA cross-linked protein polymer-coated electrodes³¹. The coated electrodes showed different cyclic voltammograms according to the size and morphology of the pores in the surface polymer film coating, which was mainly caused by the transmission of electroactive substances on the electrode interface in the polymer film pore channel or pore diffusion. Among them, the performance of macroscopic electrochemistry must take into account the size of the nanoelectrode formed by each pore in the polymer coating, the pore morphology, the distance between pores, the diffusion coefficient of the electroactive substances in the solution, and the scan rate of the electrochemical test³⁰. Generally speaking, at a faster electrochemical scan rate, the diffusion of the electroactive substances in each hole in the polymer membrane is often independent and dominant, so the electrochemical behavior of the electrode scan presents a planar diffusion curve; at a slower scan rate, the behavior of the electroactive substances at the electrode interface is dominated by diffusion, and this diffusion layer will extend beyond the pores of the polymer, so the cyclic voltammogram of the electrode presents an S shape, and its response curve shows a semi-infinite radial diffusion model, which is due to the comprehensive superposition of the S-shaped voltammograms of all pore electrodes on the total current. At very slow scan rates, the diffusion-dominant effect caused, or the distance between the pores of the polymer membrane is small enough, the diffusion layer of each micro-nanopore channel electrode will overlap with the adjacent diffusion layer, which may produce a semi-infinite linear diffusion curve, and the cyclic voltammogram of the final electrode measurement presents a typical transient voltammogram, which is a comprehensive manifestation of the polymer membrane pore electrode. In the cyclic voltammetry test experiment of the constructed BSA/g-C₃N₄/Bi₂WO₆/GA polymer film, we observed that under a series of high-speed scans, the electrode showed a transient voltammogram. According to the above hypothesis, the results inferred can show that this is due to the short distance between the pores in the coating, which leads to the overlap of the diffusion curves. By setting a simple model that only considers the diffusion effect of the electroactive substance and the distance between the pores, combined with the model constructed by Guo and Lindner³², the measured results are consistent with the predicted overlap of the pore diffusion curve, indicating that the size and distance of the pores in the polymer film we constructed are smaller than the distance of diffusion of the electroactive substance.

Although the BSA/Bi₂WO₆ coated electrode showed some redox current in the test, the BSA in the BSA/Bi₂WO₆/GA polymer coating produced dense pores due to GA cross-linking, which also increased the transmission efficiency of electroactive ions in the electrode. The electrochemical performance of the pure BSA bio-based polymer film modified electrode is also inferior to that of the electrode with the addition of a two-dimensional layered conductive material polymer coating, because the latter produces a larger current and a narrower potential difference. In order to gain a deeper understanding of this behavior, we propose a hypothesis for further research: BSA molecules at a given concentration are dispersed and adsorbed on the surface of Bi₂WO₆ loaded on the electrode. The BSA adsorption layer formed by this behavior almost eliminates the transfer of electrons to the electrode surface. The introduction of other conductive materials cannot solve this inert behavior. Even the introduction of a certain amount of GA, in materials with low polymerization degrees, also produces a situation of blocking the effective pore structure of Bi₂WO₆ particles, but when the polymer forms a film layer with a continuous shape, these Bi₂WO₆ particles are exposed and located at the far end of the thin organic coating. The charge transfer between the electroactive substance in the solution and the electrode can occur on the Bi₂WO₆ surface through the electron transfer mediated by the Bi₂WO₆ particles, as tested and observed by the electrochemical behavior. Unlike pure BSA protein cross-linked polymers that rely only on polymer pores as channels for electron transmission, the introduction of conductive two-dimensional sheet g-C₃N₄ materials can not only enhance the conductivity of the overall polymer matrix, but also the g-C₃N₄ materials rich in primary and secondary amines can serve as effective sites for urea-formaldehyde condensation reactions, and can act as effective cross-linkers for 3D polymer skeletons³³. At the same time, the dense pores formed by the polymerized materials also become ion transmission channels. Therefore, although some of the polymerized materials cover and fill the significant pores of Bi₂WO₆, the close fit between the conductive polymer matrix and the Bi₂WO₆ particles will not affect the electron transmission function, nor will it passivate the electrode.

It can be found that the polymer film formed by the mixture of g-C₃N₄ or NH₂-rGO and BSA under glutaraldehyde crosslinking also shows similar electron transfer kinetics to the BSA protein polymer, indicating that the introduction of two-dimensional conductive materials can construct ion channels and enhance the ability of the modified electrode on the flower-like Bi₂WO₆ surface to transport ions. Considering the anti-fouling ability of the polymer matrix and the detection ability of its modified electrode to heavy metal ions, it can be found that the two-dimensional material g-C₃N₄ doped with BSA polymer matrix coated Bi₂WO₆ modified electrode has stronger detection and analysis ability for heavy metal ions than graphene polymer. By comparing the formation mode of the polymer film, we speculate that it may be because g-C₃N₄ may rely on some N atoms to participate in the polymerization process during the crosslinking polymerization process, and the remaining N atoms with lone pairs of electrons can serve as the action sites for the chelation and pairing of heavy metal ions³⁴. This effect is due to the two-dimensional network of g-C₃N₄ composed of the N-rich melamine molecular structure. As can be seen from Fig. 2, with the increase in the degree of polymerization of BSA and g-C₃N₄, the proportion of Bi₂WO₆ active sites exposed increases, and the modified electrode has a stronger response ability to the electroactive substances.

Antifouling mechanism of cross-linked polymer conductive membrane

The antifouling ability of BSA cross-linked polymer coatings for macromolecules such as biological proteins is due to the fine porosity of the cross-linked polymer matrix, which can effectively prevent large biomacromolecules from diffusing through the pores. In order to further prevent the nanoscale pores from being blocked and contaminated by small molecule compounds, and at the same time, realize the migration and analysis of heavy metal ions in the pores, we developed BSA/g-C₃N₄/GA and BSA/NH₂-rGO/GA coatings. These coatings are specifically designed to match the pore size of the polymer membrane, allowing ion migration while blocking these species from reaching the electrode surface. Previous literature reports indicate that the pore diameter on the coating surface is smaller than the typical effective size of HSA protein under physiological conditions, measured at 2.74 ± 0.35 nm³⁵. The introduction of a smaller molecular weight cross-linking monomer g-C₃N₄ can more effectively regulate and generate smaller pores. In the detection of heavy metals in body fluids (blood, tears, urine, sweat, etc.) and environmental water sample (drinking water, agricultural irrigation water, groundwater, etc.), through this mode of action, due to size exclusion, the nonspecific adsorption of soluble albumin molecules (albumin accounts for about 60% of total plasma proteins) present in body fluid detection and dissolved organic matter (DOM) in water samples is prevented (Supplementary Fig. 3), so we used this abundant protein to examine the antifouling effect, which is usually the main cause of clinical POCT sensor contamination. To further explore this, we conducted a control study to determine whether proteins in human tissue fluid may be adsorbed to the coating through forces such as electrostatics and block the pores for ion transfer. 10 mg/mL HSA was incubated on BSA/Bi₂WO₆, BSA/NH₂-rGO, BSA/g-C₃N₄, BSA/NH₂-rGO/GA, BSA/Bi₂WO₆/NH₂-rGO/GA, BSA/g-C₃N₄/GA and BSA/Bi₂WO₆/g-C₃N₄/GA modified electrodes to investigate the state of electron transfer kinetics before and after incubation. It can be observed that after a long period of HSA incubation, a strong current and low ΔE_p were maintained on the BSA/Bi₂WO₆/g-C₃N₄/GA electrode.

By fixing the g-C₃N₄ content and adjusting the ratio of BSA to GA, the effects of polymer films formed at different cross-linking degrees on sensor analysis and anti-fouling performance were explored. Since g-C₃N₄ and Bi₂WO₆ are functional materials with photoelectric generation effects, the combined factors of the amount of g-C₃N₄ cross-linking in the polymer film and the amount of Bi₂WO₆ light exposure determine the photocurrent intensity. In addition, another factor that enhances the photocurrent is the Z-type heterojunction formed between Bi₂WO₆ and g-C₃N₄ molecules³⁶. This structure can enable the composite material to generate photogenerated electrons with higher efficiency in visible light absorption, and the photogenerated electrons have excellent transmission properties in the formed 3D conductive matrix³⁷. When the cross-linked polymer matrix of g-C₃N₄ and BSA is densely coated on Bi₂WO₆, it can not only promote the generation of photogenerated electrons, but also accelerate the conduction of electrons to the electrode surface and reduce the energy consumption during electron transfer (Fig. 3a). When the coatings with different polymerization degrees are contaminated by HSA biomacromolecules, the low densites or excessively high degrees of crosslinked polymer coating electrode cannot play a good anti-fouling effect, causing the generation of photogenerated electrons and the transfer effect to the electrode to deteriorate. At the same time, after incubation with HSA, its electrochemical behavior also shows different effects with the change of polymerization degree.

Fig. 3: Impact of GA cross-linking on sensor performance.

a Optimization and characterization of BSA/Bi₂WO₆/g-C₃N₄/GA coatings with varied formulations (n = 5). White bars indicate the average current density of fresh coatings, gray bars indicate 1 day incubation in 10 mg/mL HSA, and the error bars represent the standard deviation (SD) of the measurements. b Average BSA binding per coating was calculated for BSA (0.1–1 mg/mL) and GA (0.1–10%). Dashed lines (2–50) indicate the optimal range for forming 3D BSA/g-C₃N₄ matrices without excess GA. c Comparison of the deposition performance of polymer-coated electrodes with different ratios for heavy metal ions (Cd²⁺, Pb²⁺, Hg²⁺, Zn²⁺). d Investigation of the optimal deposition time of the coated electrodes produced under the optimal polymer ratio conditions for heavy metal ions (Cd²⁺, Pb²⁺, Hg²⁺, Zn²⁺).

Previous studies have shown that BSA crosslinked with GA relies on residual functional groups, including amines, phenols, thiols, and imidazoles, present in various internal amino acids³¹. Crosslinking reactions primarily involve the exposed lysine residues. Under the condition that the polymer monomer is rich in primary amines, the five-carbon GA aldehyde group will react with it quickly to form a polymer with a pyridine ring structure, which can be used as a cross-linking structural adhesive. This rapid cross-linking mechanism of dehydration condensation can form a 3D polymer molecular network and retain the native conformation of the protein polymer monomer³¹. The formation of pyridine polymers requires at least five GA molecules to produce effective crosslinks between two amine functional groups. Glutaraldehyde can effectively connect g-C₃N₄ to BSA molecules; since the proportion of g-C₃N₄ in the prepolymer is fixed, the calculation of the average cross-link number of BSA molecules can predict and evaluate the morphology of the final polymer cross-linking product. Since the BSA polymer backbone plays a major antifouling role, we mainly consider the effect of different degrees of BSA crosslinking on the performance of the polymer film under different BSA and glutaraldehyde ratios with a constant proportion of g-C₃N₄ in the prepolymer. Complexes with fewer than two intra- or inter-molecular linkages between g-C₃N₄ and BSA tend to form linear polymer chains, dimers, or other oligomeric species. In contrast, when the average number of linkages meets or exceeds the number of reactive amino acids available within the BSA molecule, the complexes form highly cross-linked and dense polymers. This cross-linking process, however, leaves an excess of unreacted aldehyde groups on the surface. There are still too many unreacted glutaraldehyde groups remaining on these highly cross-linked polymer coatings. These very active functional groups will also react with non-target detection proteins, active organic compounds, etc., in the sample to be tested, which may cause blockage and contamination of the polymer pores, resulting in poor anti-fouling ability of the polymer membrane. To optimize electrode performance, we tested several formulas with varying degrees of cross-linking. Quantitative analysis of the average number of connections per BSA molecule revealed that polymers formed by formulas such as 1 BSA/0.2 g-C₃N₄/1 GA, 0.5 BSA/0.2 g-C₃N₄/0.1 GA, 0.5 BSA/0.2 g-C₃N₄/0.5 GA, 0.1 BSA/0.2 g-C₃N₄/0.5 GA, and 0.1 BSA/0.2 g-C₃N₄/0.1 GA were predominantly composed of monomers and dimers. In contrast, formulas like 1 BSA/0.2 g-C₃N₄/10 GA and 0.5 BSA/0.2 g-C₃N₄/5 GA produced highly cross-linked coatings (Fig. 3b). Specifically, based on the stoichiometric calculation of reactive groups (-NH₂ in BSA/g-C₃N₄ and -CHO in GA) assuming a 1:1 reaction ratio for Schiff base formation, the 1 BSA/0.2 g-C₃N₄/5 GA and 0.1 BSA/0.2 g-C₃N₄/1 GA coatings are theoretically estimated to form an average of 17 and 41 cross-links between BSA and g-C₃N₄ molecules, respectively.

In this case, BSA and the conductive two-dimensional layered g-C₃N₄ molecules together form a 3D protein polymer matrix. Although this coating shows the best antifouling performance, it is not the same as the single BSA protein polymer matrix; this may be due to the effect of g-C3N4 on the change of the conductivity of the polymer membrane interface and pores. In this case, the polymerization participation and the effect of the introduced g-C₃N₄ need to be considered. In general, when a certain amount of g-C₃N₄ is introduced, the polymer has a better antifouling effect when it is in a 3D cross-linked state. In addition, a good 3D conductive coating is expected to be uniformly cross-linked, with well-defined pore sizes and a controlled number of connections between polymerized monomer molecules. We use the photocurrent density generation efficiency to further investigate the synergistic antifouling of g-C₃N₄ and BSA and its conductive efficiency. Due to the excellent electron generation efficiency of Bi₂WO₆ under visible light excitation, we can closely link the density of dense ion channels formed by BSA and g-C₃N₄ polymer films on the surface of Bi₂WO₆ and the degree of cross-linking participation of g-C₃N₄ with the photoelectron generation efficiency. 80% of the BSA protein molecules in the 0.1 BSA/0.2 g-C₃N₄/1 GA complex bind to 5–10 proteins or g-C₃N₄. In contrast, the 1 BSA/0.2 g-C₃N₄/5 GA coating has a higher average number of bonds per BSA, resulting in a more uniform and denser coating with smaller pores. This also explains why the 1 BSA/0.2 g-C₃N₄/5 GA coating has stronger anti-fouling ability and higher photocurrent density. The four toxic heavy metals of focus, mercury, lead, cadmium, and zinc, were used to further investigate the analytical performance of the above-mentioned developed electrodes (Fig. 3c). In summary, the optimal BSA/Bi₂WO₆/g-C₃N₄/GA containing 0.2 mg/mL g-C₃N₄, 1 mg/ml BSA, 5% GA and ≥1 mg/mL Bi₂WO₆ was used for further research and use, which took into account the highest current density, photocurrent generation efficiency, deposition effect of heavy metals and the lowest performance degradation after incubation in 10 mg/mL HSA for 1 d. It can be found that on this densely porous conductive polymer film, the four heavy metals can achieve the optimal deposition effect in a short time of 90 s (Fig. 3d).

Application of heavy metal detection analysis in complex matrices

Understanding the performance of sensors in complex media is critical for the field of sensing analysis for environmental monitoring and clinical diagnostics. Therefore, we conducted tolerance tests on the electrodes modified with the optimized BSA/Bi₂WO₆/g-C₃N₄/GA composites. The electrodes were exposed to 10 mg/mL HSA, untreated human serum or plasma, and wastewater for various time periods to evaluate their stability. Remarkably, after one month of exposure to various complex samples, the electrochemical sensor modified with the BSA/Bi₂WO₆/g-C₃N₄/GA coating demonstrates excellent antifouling performance, with a sensitivity loss of only ~8.5% (as shown in Fig. 4a). While the BSA/Bi₂WO₆/NH₂-rGO/GA-coated electrode retained up to 70% of its current density after soaking in 10 mg/mL HSA for 1 h, its sensitivity declined significantly with extended exposure. In contrast, the BSA/Bi₂WO₆/g-C₃N₄/GA-coated electrode maintained a high current density even after one month of immersion in complex fluids, comparable to that of pristine, unmodified gold electrodes (Fig. 4a).

Fig. 4: Antifouling performance and sensing performance of the coating.

a Comparison of the average current density recorded for BSA/Bi₂WO₆/NH₂-rGO/GA in human serum or human plasma stored at 4 °C for >30 days and BSA/Bi₂WO₆/g-C₃N₄/GA coated electrodes stored in human serum, human plasma or wastewater at 4 °C for >30 days. The data represent (n = 5) independent electrodes, with error bars showing the mean ± SD). b Changes in electrochemical impedance of the sensor fabricated using the optimal BSA/Bi₂WO₆/g-C₃N₄/GA ratio after deposition of different types of heavy metal ions (measured in PBS). c Impedance change graph extracted from the electrical impedance fitting in (b). d The sensor with the optimal BSA/Bi₂WO₆/g-C₃N₄/GA ratio was used to measure changes in the electrode’s photoelectric properties after deposition of heavy metal ions (Cd²⁺, Pb²⁺, Hg²⁺, Zn²⁺).

Using the optimized BSA/Bi₂WO₆/g-C₃N₄/GA-coated electrode, the deposition mechanism of heavy metals on the electrode surface involves a synergistic process of ion transport, electrochemical reduction, and alloy formation, which can be tracked through electrochemical impedance spectroscopy (EIS) and photocurrent measurements (Fig. 4b–d). EIS characterizes changes in charge transfer resistance (R_ct) induced by surface property alterations of Bi₂WO₆ exposed in the nanoscale ion channels after electroreduction, while photocurrent measurements reflect the shielding effect caused by heavy metal deposition passivating the active sites of Bi₂WO₆.

Stripping voltammetry is a well-established analytical method for the electrochemical detection of heavy metals, where deposition potential and solution pH are key factors influencing detection performance. By selecting a wide range of pH values and different deposition potentials, we investigated the analytical performance of the developed and optimized sensor for these four heavy metal ions. pH variation regulates the solubility of metal ions and the surface charge of Bi₂WO₆ materials, indirectly influencing deposition efficiency. The deposition potential directly controls the completeness of metal reduction and alloy formation. Collectively, these factors determine the content of electrochemically strippable metal species. The results show that the electrode exhibits the best analytical performance for these metals at pH 8.0 and a deposition potential of −1.6 V (Fig. 5a, b). At the same time, we also observed a linear relationship between concentration and response for these four heavy metal ions when mixed in equal proportions at varying concentrations (Fig. 5d, e) and when the concentration of a single ion changed. The experimental results show that there is an excellent linear relationship between concentration and response. When we use a mixed solution of four heavy metal concentrations at a specific concentration for detection and analysis, its performance remains excellent under long-term (n = 30 day) repeated (n = 5) detection. Batch-to-batch variability in electrode fabrication was evaluated by comparing the analytical performance of sensors prepared in six independent batches. All batches demonstrated consistent detection performance for the four target ions, with <5% variation in sensitivity (RSD, n = 5 measurements per batch) (Fig. 5c, f). When the concentration of Hg²⁺ is changed, the response of Cd²⁺ and Pb²⁺ will change slightly. The possible reason is that the high concentration of Hg²⁺ deposition will promote the co-deposition of Cd²⁺ to a limited extent and will slightly weaken the co-deposition effect of Pb²⁺ (as shown in Fig. 6). This situation can be ignored when detecting low concentration ions. Based on Eq. (3), the detection limits of the constructed sensor for the four ions Cd²⁺, Pb²⁺, Hg²⁺, and Zn²⁺ were 0.09 μg/L, 0.019 μg/L, 0.03 μg/L, and 0.045 μg/L, respectively. Although the BSA/Bi₂WO₆/g-C₃N₄/GA coated electrode maintains good analytical performance, the sensor’s shelf life may be limited by the composite film’s resistance to fouling under prolonged exposure to biological contaminants and use. To validate sensor performance in real samples, two authentic water samples (river water and lake water) were analyzed. Following 10 min sedimentation, supernatant aliquots were spiked with target heavy metal ions (Cd²⁺, Pb²⁺, Hg²⁺, Zn²⁺) for comparative analysis. Measurements from the fabricated electrochemical sensor showed strong agreement with ICP–MS reference data (Supplementary Table 1). Therefore, we will apply it to various real samples such as water bodies and blood (Supplementary Fig. 4). Supplementary Table 2 benchmarks state-of-the-art electrochemical platforms for heavy metal ion (HMI) quantification employing functionalized electrodes. The engineered BSA/Bi₂WO₆/g-C₃N₄/GA sensing interface demonstrates dual advantages versus literature systems: (i) enhanced sensitivity and (ii) exceptional fouling resistance in complex matrices. This synergistic performance enables direct field deployment of HMI monitoring without preprocessing.

Fig. 5: Analytical performance and stability of the sensor.

a SWV response spectra of the BSA/Bi₂OW₆/g-C₃N₄/GA coating electrode with the best ratio optimization to four heavy metal concentrations of 10 μg/L under different pH conditions. b SWV response spectra of the BSA/Bi₂OW₆/g-C₃N₄/GA coating electrode with the best ratio optimization at different deposition potentials to four heavy metal concentrations of 10 μg/L. c Performance stability of the same electrode for four heavy metal ions at fixed concentrations of 5 μg/L each under long-term use conditions. d The concentrations of Cd²⁺, Pb²⁺, Hg²⁺, and Zn²⁺ were systematically and synchronously varied from 0.05 to 10 μg/L. e The figure shows the current response extracted from d and the corresponding data of the four heavy metal concentration changes and the linear fit. Five experiments were averaged using five independent electrodes. The dots represent the mean of the current, and the error bars are the standard deviation of the mean. f Stability performance test of different electrodes via multiple current responses to four heavy metal ions, tested five times at a fixed concentration of 5 μg/L.

Fig. 6: Effect of changes in single ions in a mixture on detection.

a The concentrations of Pb²⁺, Hg²⁺, and Zn²⁺ were fixed at 10 μg/L, while the Cd²⁺ concentration was varied systematically from 0.05 to 10 μg/L. The inset shows a linear correlation between the current response and Cd²⁺ concentration within this range. b The concentrations of Cd²⁺, Hg²⁺, and Zn²⁺ were consistently fixed at 10 μg/L, while the Pb²⁺ concentration was adjusted across a range of 0.05 to 10 μg/L. The inset illustrates a clear linear relationship between the current response and Pb²⁺ concentration over the investigated range. c The concentrations of Cd²⁺, Pb²⁺, and Zn²⁺ were maintained at 10 μg/L, while the Hg²⁺ concentration was varied from 0.05 to 10 μg/L. The inset shows the relationship between current response and Hg²⁺ concentration, with changes in Hg²⁺, Pb²⁺, and Cd²⁺ concentrations influencing the responses. The data were fitted to an isothermal adsorption curve. d The concentrations of Cd²⁺, Pb²⁺, and Hg²⁺ were fixed at 10 μg/L, while the Zn²⁺ concentration was systematically varied from 0.05 to 10 μg/L. The inset highlights a linear correlation between the current response and Zn²⁺ concentration within the tested range.

Discussion

In this study, we developed a binary condensation conductive bio-based polymer membrane that demonstrates excellent antifouling and conductive properties on the surface of the working electrode. When combined with bismuth-based materials with a high specific surface area, the composite coating further enhances the electrochemical sensor’s performance in heavy metal analysis. The preparation of the polymer membrane is based on the antifouling properties of BSA molecules for various large-sized biomolecules, the introduction of highly conductive and N-rich two-dimensional g-C₃N₄, and the effective regulation of the cross-linking degree of 3D biomolecules and conductive two-dimensional layered materials on the size of ion migration channels through the urea-formaldehyde condensation reaction. This composite coating successfully alleviates the challenge of electrode material fouling, even under complex biological fluids, thereby improving the diagnostic performance of electrochemical sensors and facilitating their long-term reuse on the same chip. Due to the construction of smaller-scale ion channels within the pores achieved by cross-linking the conductive two-dimensional layered material g-C₃N₄ with BSA, the diffusion of ions and electrons is enhanced, thereby promoting the correct signal transmission on the working electrode. A key aspect of our fabrication approach lies in the one-pot in situ polymerization process. Compared with the layer-by-layer assembly method, this in-situ modification process can not only reduce the differences between chip sensor preparations, but also be continuously batched, low-cost and time-effective. This is essential for future commercialization at scale. By employing this method, we constructed 300 electrochemical sensors within 25.5 h. These sensors are capable of simultaneously detecting trace toxic heavy metal ions Cd²⁺, Pb²⁺, Hg²⁺, and Zn²⁺ with high sensitivity and selectivity (shown in Supplementary Tables 3 and 4). The outstanding antifouling properties of the composite material prevent signal attenuation caused by non-specific species in plasma, serum, and water samples. This advantage has the potential to simplify the sample pretreatment steps in field detection and analysis applications. The developed high-precision sensing (Supplementary Table 5) and analytical technology enable the rapid collection of large-scale field data through simple detection methods. This allows for a deeper understanding of the sources of heavy metal pollution and the correlations between different pollution sources. Moreover, it holds the potential for targeted, personalized detection and analysis of potential pollution sources, thereby improving detection measures and personalized pollution control strategies to meet the urgent demands of rapid emergency response to heavy metal emissions.

This composite coating modification technology holds great potential for applications in the field of electrochemical sensors. The polymer thin film coating, with its excellent antifouling properties, can effectively prevent signal interference and attenuation caused by non-target substances in the test solution under complex environmental conditions. As a result, it simplifies the sample preparation steps in field detection applications, streamlines the overall testing process, and reduces the generation of false signals. In addition, the continuous introduction of a variety of active functional polymer monomer materials into the functional construction system of porous composite materials can allow the expansion of the analytical application range of various ions and organic pollutants to be tested. This comprehensive diagnostic capability is essential for managing future heavy metal exposure diseases and tracing environmental heavy metal pollution, because rapid and accurate diagnosis is essential. Large amounts of test data can be collected efficiently and cost-effectively in a simplified manner, and this strategy should enhance our understanding of the correlation between human heavy metal exposure levels and the living environment. This will not only help improve diagnostic accuracy and monitor environmental heavy metal exposure levels, but also help us understand the extent of heavy metal pollution and customize patient management strategies for removing heavy metals from the body. Ultimately, the establishment of a robust nonspecific adsorption barrier also presents opportunities for the design of other types of sensor interfaces, including implanted devices and medical diagnostic equipment. By reducing undesirable interactions with nonspecific molecules, this technology has the potential to improve the analytical performance and lifespan of sensors, while possibly alleviating rejection reactions with the analytical receptors, thereby improving their reliability.

Methods

Synthesis of g-C₃N₄, NH₂-rGO, and flower-like Bi₂WO₆

The preparation method of g-C₃N₄ mainly involves the thermal condensation of melamine with thiourea. Similar to previous literature reports³⁸, we made a slight modification. Briefly, a solution containing melamine (4.6 g/L) and thiourea (1.25 g/L) was thoroughly mixed and sealed in a ceramic crucible. The crucible was placed in a muffle furnace, heated at a rate of 5 °C/min to 580 °C, and maintained at this temperature for 1 h. After natural cooling, yellowish g-C₃N₄ lumps were collected from the bottom of the crucible, dried at 100 °C, and ground into a fine powder for further use. The as-obtained g-C₃N₄ powder was ultrasonically dispersed in a certain amount of ultrapure water for 2 h, and the suspension was then centrifuged at 12,000 rpm to obtain ultrathin g-C₃N₄ nanosheets.

The single-layer rGO was purchased through Jiangsu Xianfeng Nanotechnology Co., Ltd. NH₂-rGO preparation: We followed the method reported in previous literature and made slight modifications to the method. Add 0.12 g of single-layer rGO to 40 mL of ethylene glycol and disperse under ultrasonic treatment. Further add 1 mL of ammonia water, transfer the dark brown solution to a high-pressure vessel lined with Teflon, and place it at 180 °C for solvent thermal reaction for 10 h. After the reaction is complete, filter the reactants. Leave the precipitate behind, wash it repeatedly with distilled water, and place it in a 60 °C drying oven for 24 h. After drying, further use. Unless further mentioned, the samples used in the following electrochemical measurements were prepared under these conditions.

We modified the method reported in the literature³⁹ and synthesized flower-like Bi₂WO₆ via coprecipitation. Bismuth nitrate (Bi(NO₃)₃) and sodium tungstate (Na₂WO₄) were used as precursors. Specifically, 0.919 g Bi(NO₃)₃·5H₂O and 0.833 g Na₂WO₄ were dissolved in 40 mL of distilled water, stirred at room temperature for 3 h, and ultrasonicated for 5 min every hour. The mixture was then transferred to a sealed 50 mL Teflon-lined crucible and heated at 150 °C for 20 h. Finally, the resulting precipitate was collected, thoroughly washed alternately with ultrapure water and anhydrous ethanol, and dried at 75 °C for 7 h.

Preparation of BSA/Bi₂WO₆/g-C₃N₄/GA cross-linked complex conductive polymer film

The g-C₃N₄ nanosheets, flower-shaped Bi₂WO₆, BSA and glutaraldehyde solution prepared above were mixed in a certain proportion. Specifically, 2.95 mL of g-C₃N₄/Bi₂WO₆ (with varying proportions, where the concentration of g-C₃N₄ and Bi₂WO₆ was ≥1 mg/mL) was mixed with 5–500 μL 70% GA. The mixture was shaken for 10 min and then ultrasonicated for 1 min, yielding a cross-linked prepolymer of BSA, g-C₃N₄ and glutaraldehyde wrapped with flower-shaped Bi₂WO₆ encapsulated within.

Fabrication of BSA/Bi₂WO₆/g-C₃N₄/GA/SPE sensor

After each grinding of the working electrode, clean the electrode with ethanol and ultrapure water under ultrasound. Prepare the composite solution according to the optimized formulation (1 mg/mL BSA/0.2 mg/mL g-C₃N₄/5 mg/mL GA) and mix thoroughly using a vortex mixer. Subsequently, rapidly pipette 70 µL of the pre-polymerized solution onto the surface of the working electrode and allow it to dry. Then place the electrode in a water-moistened box, and incubate the box in a constant-temperature shaker at 25 °C for 24 h to ensure sufficient polymerization. Upon reaction completion, the electrodes were dried in an oven at 40 °C for 60 min. Finally, it was rinsed with ultrapure water and ethanol to remove unreacted compounds, reactive monomers and oligomers. The prepared electrode sensor was used for the next experimental steps.

Electrochemical characterization of polymer coating modified electrodes

The planar microfabricated chip used in the electrochemical experiments contains a gold working electrode (approximately 1.5 mm²), a gold reference electrode, and a gold counter electrode. The sensor was connected to an IGS6030 electrochemical workstation (Guangzhou Ingsens sensor technology, China) via a homemade connection box and controlled by IGS4030CN software for data acquisition. During the coating preparation process, the percentage of GA, the concentration of g-C₃N₄ and BSA were optimized: the content of g-C₃N₄ was fixed at 0.2 mg/mL, the concentration range of BSA was 0.1, 0.5, and 1 mg/mL, and the range of GA was 0.1, 0.5, 1, and 10% (v/v). The coating time was optimized as follows: the pre-polymer mixture was drop-coat onto the working electrode surface and incubated at 25 °C for 1 day; afterward, the electrode chip was flushed with PBS buffer using a syringe (injection needle hole specification: 0.5 mm) at a flow rate of 1 mL/min for 5 min, with the distance between the chip and the needle being 1 cm. The coating was electrochemically characterized before and after incubation in 10 mg/mL HSA for 1 day. Nanocomposite coatings were characterized by cyclic voltammetry (scan rate of 100 mV/s, potential range: −0.5 and 0.5 V vs. open circuit potential) and EIS (frequency range: 0.1 Hz to 0.1 MHz, amplitude: 5 mV amplitude vs. open circuit potential, 50 logarithmic interval measurements) in an aqueous redox solution containing (5 mM K₃Fe(CN)₆/K₄Fe(CN)₆ in 1 M KCl).

Four heavy metal standard solutions (1000 mg/L) were diluted to an intermediate concentration of 1000 μg/L using analytical buffer. This intermediate solution was then serially diluted with the same buffer to generate calibration standards at target concentrations of 0.05, 0.08, 0.1, 0.2, 0.25, 0.5, 1.0, 3.0, 5.0, 7.5, and 10.0 μg/L for subsequent electrochemical analysis. The solution used for method optimization was a 10 μg/L mixed solution of four heavy metals: Cd²⁺, Pb²⁺, Hg²⁺, and Zn²⁺. During the determination using four different electrochemical methods, parameters such as deposition time and deposition potential were held constant. Following the single-variable principle, only the electrochemical voltammetry mode was altered for the determination. Heavy metal ions are measured using square wave voltammetry. The detection process consists of two consecutive stages: an open-circuit pre-concentration (deposition) step of the analyte solution on the electrode, followed by transfer of the electrode to the analytical buffer solution for voltammetric detection. The first step involves immersing the electrode chip into a beaker containing 20 mL of heavy metal solution at a specific concentration. After a predetermined deposition time under mild stirring, the electrode is quickly removed and rinsed with ultrapure water. The electrode is then transferred to an electrochemical cell containing the detection buffer. All experiments were repeated five times, with the data points representing the average of five repetitions.

BSA/Bi₂WO₆/g-C₃N₄/GA/SPE sensor fouling test

All reagents were prepared in the assay buffer, and the electrode was rinsed with 2 mL of the same buffer after each testing step. Functionalized or unfunctionalized BSA/Bi₂WO₆/g-C₃N₄/GA/ coated electrodes were incubated with HSA, human serum and plasma, or wastewater for different lengths of time, and then tested by cyclic voltammetry in an electroactive potassium ferrocyanide solution.

Determining diffusion-limited electrochemical processes by the Randles–Sevcik equation

To diagnose solution-phase diffusion control of an analyte, cyclic voltammetry responses should exhibit a linear dependence of peak current (i_p) on the square root of scan rate ($\sqrt{\nu }$). Concurrently, consistent peak separation (ΔE_p) across scan rates validates reversible behavior. This diagnostic criterion is formalized by the Randles–Sevcik equation:

$${i}_{p}=0.446n{{FAC}}^{0}{(\frac{nF\nu {D}_{0}}{RT})}^{1/2}$$

(1)

where n denotes electron count per redox reaction, A (cm²) corresponds to the electrode’s geometric area, D₀ (cm² s⁻¹) represents the analyte’s diffusion coefficient, and C₀ (mol cm⁻³) signifies its bulk concentration.

Calibration of standard curve, LOD, and LOQ

The calibration standard curve of the sensor was constructed by measuring the oxidation current intensity of four heavy metal ions at different concentrations under mixed conditions, and fitting the data to a linear regression:

$$y=a\times x+b$$

(2)

Here, y represents the current intensity, and x denotes the analytical concentration of the ion being measured. a is the slope and b is the intercept, which are the parameters obtained by fitting the linear. The limit of detection (LOD) and limit of quantification (LOQ) we defined are the concentration on the calibration curve corresponding to y_LOD and y_LOQ, respectively, where y_LOD is the average current of the blank sample plus three times the standard deviation of the blank response, denoted as y_LOD:

$${y}_{{{\rm{LOD}}}}=\frac{\overline{{y}_{{blank}}}+3{\sigma }_{{blank}}-b}{a}$$

(3)

$${y}_{{LOQ}}=\frac{10{\sigma }_{{blank}}}{a}$$

(4)