Flexible-fabricated sensor module with programmable magnetic actuators coupled to L-cysteine functionalized Ag@Fe3O4 complexes for Cu2+ detection in fish tissues

Heavy metal contamination for seafood, particularly fish, is arising great concerns, and consequentially it is necessary to develop a simple and direct detection method. In this work, Ag@Fe3O4 is successfully prepared by simple solvothermal method, and we present a flexible-fabricated sensor module with assembled programmable magnetic actuators. The resulting sensor integrates a three-electrode system with two programmable magnetic actuators at the bottom of the device, which regulates the amount of current by adjusting the brake to control the adsorption force and vibration. The L-Cysteine functionalized Ag@Fe3O4 is coated on the surface of the electrode, then the Cu2+ is dropped into the reaction tank. Its performance is studied by cyclic voltammetry and electrochemical impedance spectroscopy, and the key experimental conditions such as deposition potential, deposition time, and electrolyte pH are gradually optimized. Under optimal conditions, Cu2+ can be detected over a wide linear range (0.01 ~ 4 μM) and at a low LOD (0.34 nM). The results show that the proposed method has a good application prospect in the detection of Cu2+. This method is successfully applied to Cu2+ analysis in fish samples with an acceptable recovery of 93 ~ 102%.


Introduction
Fish is rich in vitamins, proteins and other valuable nutrients, which are easily digested and absorbed by human body, therefore it is considered to be one of the most important food sources for global consumption (Khalili Tilami and Sampels 2018; Idowu et al. 2021). However, fish are frequently contaminated with heavy metals from sources such as domestic sewage and industrial wastewater (Shahjahan et al. 2022;Dallinger et al. 1987). As a representative heavy metal, Cu 2+ in fish can be used as a reliable indicator of environmental pollution of water sources (Javed and Usmani 2019;Song et al. 2015). The concentration level of Cu 2+ in human body directly affects human health, when Cu 2+ content of the human body accumulates to a certain extent, it will cause serious harm to life and health, resulting in biochemical disorders, physiological dysfunction and various pathological changes in internal organs (Yu et al. 2021).
Various traditional detection techniques, such as inductively coupled plasma atomic emission spectrometry (ICP-AES) (Ebrahimi-Najafabadi et al. 2019), inductively coupled plasma mass spectrometry (ICP-MS) (Suo et al. 2019), and atomic absorption spectrometry (Ferreira et al. 2018) have been used to detect Cu 2+ . However, these technologies are often complex, time-consuming, relatively expensive, and require bulky and non-portable devices. Electrochemical sensor has been recognized as a promising powerful tool for the determination of heavy metal ions due to its simplicity, low cost, portability, and high sensitivity. In electrochemical analysis, the accumulation of targets on the electrode surface facilitates the sensitivity of the response signal and the sensor (Jothimuthu et al. 2011;Shaik et al. 2021). Among them, magnetic separation is considered as an efficient separation-pre-concentration technology, which has the advantages of simple operation, high enrichment coefficient and wide application range (Kim et al. 2009a).
According to previous studies, the integration of magnetic nanoparticles (MNP) into sensors can improve the performance of the sensors (Rocha-Santos 2014). Fe 3 O 4 magnetic nanoparticles have been widely used in the manufacture of biosensors due to their superparamagnetic, large specific area, and ease of preparation, which can be readily separated from the reaction mixture with a magnet and weighed immediately after removing the magnet new dispersion (Yuan et al. 2017;Guo et al. 2020). However, Fe 3 O 4 still suffers from problems such as easy agglomeration of particles and oxidation in the process of application (Freitas et al. 2016). One-dimensional nanomaterials (wires, rods and tubes) have attracted widespread attention in electrochemical sensors due to their unique electrical, catalytic, physical, and chemical properties (Bi and Lu 2008). Nanowire-based biosensors exhibit improved signal-to-noise ratios, fast electron transfer rates, enhanced sensitivity due to the rapid electron transfer between electrode and analyte (Qin et al. 2011). The stability of Fe 3 O 4 under corrosive conditions was enhanced by combining silver nanowires (AgNWs) with Fe 3 O 4 .
In order to specifically identify Cu 2+ , the surface of the composite material Ag@Fe 3 O 4 nanowires (hereinafter collectively referred to as Ag@Fe 3 O 4 ) must be functionally modified (Atapour et al. 2019). L-Cysteine (L-Cys) has been an important amino acid found in natural proteins that plays a key role in biological systems (Rohilla et al. 2020). The amino nitrogen and carboxyl oxygen in the L-Cys molecule have solitary pairs of electrons and can form stable coordination compounds with Cu 2+ (Yang et al. 2001). The L-Cys molecule assembled on Ag@Fe 3 O 4 surface through the Ag-S bond (Betts et al. 2020). Traditional electrochemical sensors, such as microfabrication techniques, were extensively used in semiconductor industry due to its satisfactory flexibility and outstanding control capability in sensors parameters. But manufacturing these sensors may require a variety of technologies and specialized facilities such as lithography and sputtering (Shamkhalichenar et al. 2020). As an alternative, flexible printed circuit technology offers various advantages such as flexibility, low manufacturing temperatures, and material savings (Li et al. 2020).
In this study, a facile polyol synthesis route is used to synthesize AgNWs by the reduction of AgNO 3 with ethylene glycol, and then the Ag@Fe 3 O 4 nanocomposites are formed by loading Fe 3 O 4 on silver nanowires by the solvothermal method. An analytical method for enriching Cu 2+ traces using L-Cys functionalized Ag@Fe 3 O 4 nanocomposites is established, and L-Cys assembled on the surface of Ag@Fe 3 O 4 through Ag-S bonds is able to capture Cu 2+ . Flexible-fabricated sensor module is prepared, based on the technique of screen-printed electrode (SPE), to form an integrated three-electrode system. Meanwhile, the sensor module integrates a pair of programmable magnetic actuators for extensive analytical facilitations. The integrated design facilitates the miniaturization and systematization of the sensor and controls the current level by adjusting programmable magnetic actuators, thereby controlling the adsorption force. Under magnetic force, the L-Cys functionalized Ag@ Fe 3 O 4 is adsorbed directly onto the surface of the working electrode, and then the Cu 2+ liquid is added, accelerating the adsorption of the Cu 2+ liquid through the alternating magnetic field generated by the assembled programmable magnetic actuators, while the programmable magnetic actuators vibrate under the action of the alternating magnetic field and can act as an alternative to stirring.

Reagents and materials
Cu 2+ standard solution (1000 μg/mL) was obtained the China Center for Standard Reference Materials. Polyvinylpyrrolidone (PVP) was purchased from Sigma-Aldrich. Hydrogen peroxide (30%), AgNO 3 (0.01 M, A26785), Fe(NO 3 ) 3 ·9H 2 O, ethylene glycol, NaOH, diethylene glycol, and other chemical reagents were obtained from Sinopharm Chemical Reagent and were analytical grade and do not require further purification. Ag/AgCl paste (AS5900) was obtained from Changxiang Industrial Co., Shanghai. Deionized water was produced using the Milli-Q system (USA) throughout the experiment.

Synthesis of L-Cys/Ag@Fe 3 O 4 composites
AgNWs was synthesized by reducing AgNO 3 with ethylene glycol by polyol method (Fahad et al. 2019). Briefly, 20 mL of ethylene glycol was heated first in an oil bath at 150 °C for 1 h, then 5 mL of AgNO 3 -ethylene glycol solution (containing 0.24 mol AgNO 3 in ethylene glycol) and 5 mL of PVP-ethylene glycol solution (containing 0.24 mol PVP in ethylene glycol) were added drop by drop. The solution became turbid and grey in about 15 min, indicating the appearance of silver nanowires, and continued to reflect at 150 °C for 5 h, leaving a grey precipitate at the end of the reaction. Finally, the solution was washed 3 ~ 5 times with ethanol and acetone and dried at 60 °C for 10 ~ 12 h.
The AgNWs (0.1 mmol) and Fe (NO 3 ) 3 (0.4 mmol) were first dissolved in 10 mL of diethylene glycol, then sodium acetate (0.01 mol) and 3 mL of ethylene glycol were added in turn to obtain an orange-brown solution. Then, resulting mixture was transferred to a 20 mL PTFE-lined stainlesssteel autoclave, which was sealed and kept at 200 °C for 10 h. The autoclave was cooled naturally to room temperature. The precipitate Ag@Fe 3 O 4 was washed 3 times with distilled water and ethanol and dried at 30 °C for 6 h. Finally, the product was dispersed in ethanol and stored at room temperature for further characterization. In addition, pure Fe 3 O 4 nanoparticles were fabricated as a comparative test.
L-Cys can bind to Ag@Fe 3 O 4 nanowires through Ag-S bonds. 10 mM L-Cys solution was prepared for experimental use. The prepared L-Cys solution and Ag@Fe 3 O 4 nanowires were first mixed in a beaker and incubated for about 30 min. The L-Cys/Ag@Fe 3 O 4 composites were collected by magnet adsorption. Finally, the composites were rinsed with deionized water to remove the incompletely bound L-Cys.

Manufacture of flexible sensors
The screen printer is purchased from Shenzhen Chuanglida Printing Machinery Co. The working principle is to print the ink on the substrate (PVC, glass fiber, ceramic, polyester film, alumina, etc.) through layer-by-layer deposition, using the screen stencil to form the electrode pattern, then the electrode is baked to remove the solvent from the ink so that the electrode strip is shaped and cured.
Sensor module was prepared based on our well-establish technique (Su et al. 2022;Zhou et al. 2023;Yuan et al. 2023) but with innovated magnetic actuator integration (Fig. 1A). Briefly, the module (length 30 mm, width 13 mm) was initially fabricated on 0.2 mm flexible polyimide substrate using flexible printed circuit technique. All pads on the top layer, as semi-finished electrodes, were made with 17 μm Cu film (XLF-075) and covered with 0.5 μm Au. The semicircular pad was directly adopted as counter electrode. The arc one was subject to Ag/AgCl paste coating to organize functional reference electrode. The central round pad (diameter = 4 mm) was adopted as working electrode until further modification. Meanwhile, a pair of E-shape Mn-Zn ferrites (Type: EE10, magnetic permeability μ > 10,000(H/m)) were commercially purchased. By assembling windings (N = 600, with 0.08 mm varnished wire) to each of them, the magnetic actuators were well obtained. These magnetic actuators were glued on the bottom of the sensor module with tiny magnetic gap as 0.5 mm, and sealed with moisture-resisted epoxy. Thus, they could accept external AC or DC excitation (0 ~ 1 V) for desired performance ( Fig. 1B and C).
Synchronously, two independent pad groups were fabricated onto the bottom of the SPE, well forming the dual electric field driver. Area distribution for each electric field unit could be trimmed by assembling bypass resistors. During operation, external excitation was applied to this programmable electric field driver and the working electrode.

Sensor modification and electrochemical measurement
Electrochemical experiments were performed using the CHI-660D analyzer (CH Instruments, China). Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) were tested for 5 mM K 3 [Fe(CN) 6 ]/K 4 [Fe(CN) 6 ] and 0.1 M KCl. In CVs, the scanning rate and potential window was 0.1 V and -0.2 ~ 0.6 V. The frequency range and amplitude of EIS were set as 10 -2 ~ 10 5 Hz, and 5 mV. The DC potential of EIS was set as 0.23 V, which was equivalent to the formal potential of the K 3 [Fe(CN) 6 ]/K 4 [Fe(CN) 6 ] redox probe . The temperature is about 25 °C, and the humidity is 35%-70% during the electrochemical measurements.
Previous studies had reported that Cu 2+ can be stably bound to the amino nitrogen and carboxyl oxygen in L-Cys. 40 μL of the prepared L-Cys/Ag@Fe 3 O 4 composite was taken and dropped uniformly onto the SPE surface, followed by 10 μL of Cu 2+ solution. After Cu 2+ bound to L-Cys, due to the good paramagnetic properties of Fe 3 O 4 , the composite material is adsorbed onto the SPE surface by means of an electromagnet assembled on the bottom of the device and incubated on the surface for 30 min (Fig. 1D). The pH was optimized to ensure a certain affinity between Cu 2+ and L-Cys. Sample pre-treatment had removed most impurities and after incubation and natural drying, the surface was rinsed with deionized water. Cu 2+ was determined by differential pulse adsorption solvation voltammetry (DPAdSV) (Fu et al. 2013). 10 mL of PBS (0.1 mol L −1 ) was dropped into the reaction tank. DPAdSV scanned the working electrode over a range of -0.4 to 0.6 V. At -0.4 V, Cu 2+ was reduced to metallic Cu.

Sample pretreatment
Fish samples were purchased from the local market, and the muscle was separated, cut into small pieces and rinsed three times with deionized water. Place 1 g of dried sample in the digestion vessel and add 10 mL of HNO 3 (70%) and 12 mL of H 2 O 2 (30%). The vessel was immediately assembled and placed in a 150 °C microwave digester for approximately 5 h. After complete digestion of the sample, the solution was cooled to room temperature and the pH of the digestion solution was adjusted to neutral, the liquid was dried under vacuum to 1 g. Finally, the resulting solid was dissolved in 1 mL of deionized water. The prepared solution was stored in a refrigerator at 4 °C in preparation for further determination.

Characterization of Ag@Fe 3 O 4
AgNWs and Ag@Fe 3 O 4 were characterized with SEM to observe the morphology of the synthesized materials. Figure 2A and B showed typical SEM images of the prepared AgNWs, and it could be seen that the AgNWs have a relatively smooth surface and with random arrangements. The length of AgNWs was in the range of 40 ~ 60 μm, and the diameter was about 200 nm. The typical diameter of Ag@ Fe 3 O 4 nanowires was about 400 nm, the length was similar to that of AgNWs, and the surface of AgNWs was coated with Fe 3 O 4 (Fig. 2C). Figure 2D showed a schematic diagram of Ag@Fe 3 O 4 subjected to magnetic adsorption. EDS spectra were employed to illustrate elemental compositions of AgNWs and Ag@Fe 3 O 4 . In Fig. 2E, Ag element was distinctly observed in AgNWs spectrum, and after the addition of Fe 3 O 4 , the content of Ag element decreased significantly and the content of C and Fe increased.

Electrochemical characterization
Electrochemical behavior of bare SPE, Ag@Fe 3 O 4 /SPE and L-Cys/Ag@Fe 3 O 4 /SPE were compared with CV in 5 mM K 3 [Fe(CN) 6 ]/K 4 [Fe(CN) 6 ] and 0.1 M KCl. Symmetrical shapes and reversible redox peaks were observed on bare SPE as shown in Fig. 3A. The modification with Ag@Fe 3 O 4 showed a significant increase in peak current, which was related to the good electrocatalytic properties of AgNWs, as their electrical properties and adsorption ability could effectively facilitate the electron transfer between the solution and electrode interfaces. The peak currents of Ag@Fe 3 O 4 /SPE were further enhanced by the addition of AgNWs. The redox peak decreased after L-Cys incubation, which was caused by cysteine insulation (Cheng et al. 2016).
EIS was considered to be an effective tool for probing the interfacial properties of surface-modified electrodes. In a typical EIS, the diameter of the semicircle is equal to the electron transfer resistance (Ret), which reflected the electron transfer kinetics of the redox probe at the electrode surface. As shown in Fig. 3B, the Ret value of Ag@Fe 3 O 4 /SPE was much higher than that of bare SPE, which was due to the fact that Fe 3 O 4 was a poorly conductive metal oxide with a large electron transfer impedance, as such while AgNWs were able to accelerate the electron transfer on the electrode surface and reduce the electron transfer resistance. After modifying AgNWs@Fe 3 O 4 with L-Cys, the Ret value of L-Cys/Ag@Fe 3 O 4 /SPE was lower than that of Ag@Fe 3 O 4 / SPE because L-Cys could provide tunable conductivity and fast electron transfer (Moccelini et al. 2008).
To explore the potential analytical benefits of benefiting from Ag@Fe 3 O 4 , bare SPE, Fe 3 O 4 /SPE and Ag@Fe 3 O 4 /SPE were used for CV analysis (Fig. 3C). Then, calculated the electroactive area of these electrodes according to the Randles-Sevcik equation (Su et al. 2022): where Ip was the peak current, A was the active area of the electrode, D was the diffusion coefficients, the value was 7.6 × 10 -6 cm 2 /s, c was the concentration of K 3 [Fe(CN) 6 ], the value was 5 × 10 -6 mol L −1 , and v was the pre-stabilized scan Ip = (2.69 × 10 5 )AD  (Fig. 3D). It illustrated that Ag@Fe 3 O 4 /SPE presented much larger electroactive area than bare SPE or Fe 3 O 4 /SPE.

Optimization of experimental parameters
For best performance, key parameters such as deposition potential, deposition time, electrolyte pH, incubation time, driving voltage, and magnetic field frequency were optimized to maximize the efficiency and sensitivity of the electrode.
The effect of deposition potential on current was studied over a potential range of -1.5 V to -1 V. As shown in Fig. 4A, with the deposition potential increases from -1.5 V to -1.2 V, the peak current gradually increases until -1.2 V. At higher potentials (-1.2 V to -1.0 V), the peak intensity of the peeling current decreases, which was caused by hydrogen reduction on the surface of the modified electrode. Therefore, the -1.2 V potential was optimal for detecting Cu 2+ . In addition, a typical raw data plot of DPAdSV for the deposition potential was shown in Fig. S1.
Controlling the deposition potential to -1.2 V, the response current raised rapidly with the deposition time between 30 ~ 120 s, and did not change much after 120 s (Fig. 4B), the deposition signal peaks within 120 s, indicating that the electrode surface had reached the maximum load capacity, and the extended deposition time led to the loss of the peel current strength. Therefore, 120 s was chosen for this experiment.
The selection of the appropriate pH was also crucial, as pH affected the sensitivity and selectivity of quantitative studies. Therefore, the voltametric signal of Cu 2+ was examined in PBS over the pH range from 3 to 9. Figure 4C showed the peak current versus electrolyte pH. The peak current increased with increasing pH from 2 to 5 and the maximum peak current was observed at pH = 5. It was possible that when the pH was small (< 5), competition and repulsion affect the binding effect of Cu 2+ to L-Cys, while when pH > 5, the interaction between the amino nitrogen in L-Cys and AgNWs was enhanced, thus weakening the linkage of the amino group to Cu 2+ .
L-Cys was assembled on the Ag@Fe 3 O 4 surface via Ag-S bond. the amount of L-Cys binding on the Ag@Fe 3 O 4 surface was dependent on the incubation time. As shown in Fig. 4D, the peak current increased significantly with the incubation time up to 30 min and maintained a stable response after 30 min. This may be due to the increase in the amount of L-Cys binding on the Ag@Fe 3 O 4 surface from 1 to 30 min during which the amount of trapped Cu 2+ also increased, saturation was reached after about 30 min.
Both the drive voltage (which provides the excitation source for the magnetic brake to operate) and the magnetic field frequency of the magnetic brake also affected the peak current, because the vibrational effect provided by the magnetic brake played an important role in the incubation process. Controlling the deposition potential (-1.2 V), deposition time (120 s), pH (7) and incubation time (30 min) for optimal optimization, the effect of drive voltage and magnetic field frequency on the peak current was investigated. As shown in Fig. 4E, when the drive voltage was set at 0 ~ 0.6 V, the peak current increased rapidly, but with higher voltages the vibrations became more intense and many bubbles form in the sample solution, impeding the transfer of mass at the sensing interface and causing the peak current to decay. Similarly, as the magnetic field frequency increased (Fig. 4F), the number of discharges per unit time increased and the peak current tended to increase rapidly, however, when the frequency was set too exorbitant, it was conjectured that the brake temperature increased due to the increase in average discharge energy per unit time, affecting the binding of L-Cys to Ag@Fe 3 O 4 .
In addition, the effect of different thickness sensors (corresponding to different flexibilities) on the peak current was studied in the range of 0.1-0.3 mm. The thickness of the sensor is proportional to the thickness of the copper, as shown in Fig. S2. When the sensor thickness was between 0.1 and 0.2 mm, the peak current gradually increases, peaking at 0.2 mm and then gradually decreasing. Therefore, a 0.2 mm thickness sensor was selected.

L-Cys/Ag@Fe3O4/SPE performance analysis
The analytical performance of L-Cys/Ag@Fe 3 O 4 /SPE for Cu 2+ detection was investigated under optimized conditions. As shown in Fig. 5A and B, the concentration of Cu 2+ was linear in the range of 0.01 ~ 4 μM. The linear regression equation for the analytical current with Cu 2+ concentration was Ip (μA) = 0.064C (μM) + 2.169 (R 2 = 0.992), where Ip represented the peak current and C represented the concentration of Cu 2+ (mol L −1 ). The sensitivity for Cu 2+ detection was 0.064 μA/(μM), and the limit of detection (LOD) was 0.34 nM. LOD was defined as LOD = 3Sb/S, where Sb and S were standard deviation of 10 blank signals and the slope of the calibration curve. Compared with previously reported electrochemical electrodes for Cu 2+ , it could be seen that the proposed L-Cys/Ag@Fe 3 O 4 /SPE presented low detection limit, high sensitivity, and simple synthesis (Table 1).

Interference, repeatability, long-term stability, and reproducibility
A number of interfering metal ions were detected under the same optimized conditions, including 10 μM (10 folds of Cu 2+ ) Zn 2+ , Pb 2+ , Mg 2+ , K + , Na + , Ca 2+ , Hg 2+ , Cd 2+ , and Ni 2+ . As shown in Fig. 5C, the responses of these interfering ions were very small at such high concentrations. In order to and Cd 2+ (100 folds of Cu 2+ ) were investigated (Fig. 5D), and the responses of these interfering ions remained small at such high concentrations. Therefore, it indicated that these ions do not interfere with the determination of Cu 2+ . This is because Cu 2+ has stronger coordination with L-Cys compared to other metal ions and forms stable complexes under the same conditions. The reproducibility and stability of L-Cys/Ag@Fe 3 O 4 / SPE were investigated by continuous cycling electrodes. As shown in Fig. 5E, the relative standard deviation (RSD) is 2.360%, which confirmed L-Cys/Ag@Fe 3 O 4 /SPE was stable and reproducible for Cu 2+ detection.
Storage stability was studied by measuring the DPADSV response to Cu 2+ every two days for 15 days, and compared the peak current on the first and last days, it was found to decrease slowly, from 3.6 μA on the first day to 3.16 μA, maintaining 88% of the initial response. It was shown that L-Cys/Ag@ Fe 3 O 4 /SPE has good storage stability (Fig. 5F).
In order to achieve experimental reproducibility, five sensors of L-Cys/Ag@Fe 3 O 4 /SPE were prepared using the same method, as shown in Fig. S3, corresponding to five sets of peak current magnitudes around 3.5 μA, demonstrating the good reproducibility of L-Cys/Ag@Fe 3 O 4 /SPE.

Electrochemical detections of Cu 2+ in fish
The performance of L-Cys/Ag@Fe 3 O 4 /SPE for practical applications was evaluated by analyzing fish samples. To verify the accuracy and reliability of the method, three sets of breams, grass carp and black carp sample extracts were prepared for DPAdSV and ICP-MS based assays, respectively. As shown in Table 2, the results of DPAdSV assay were compared with those of ICP-MS with error ratio in the range of 3.7 ~ 9.3%. In addition, 0.5 μM, 1 μM and 4 μM of Cu 2+ were added to three groups of fish samples and analyzed using DPAdSV with recoveries in the range of 93 ~ 102% and relative standard deviations (RSD) < 6.6%. These results indicated that L-Cys/ Ag@Fe 3 O 4 /SPE was suitable for Cu 2+ determination with excellent accuracy, precision and reliability in the analysis of natural samples, and showed good applicability for the determination of real samples.

Conclusion
In this study, we successfully synthesize Ag@Fe 3 O 4 composites using a solvothermal method, establish an analytical method for the trace enrichment of Cu 2+ using L-Cys-functionalized Ag@Fe 3 O 4 composites. Furthermore, we introduce a flexiblefabricated sensor module with assembled programmable magnetic actuators for the differential pulse adsorption solvation voltammetry of Cu 2+ in fish samples. Compared with previously reported methods for electrochemical determination of Cu 2+ , this method has the advantages of simple fabrication, low detection limit (0.34 nM), excellent anti-interference, and low cost inexpensive. Under optimal conditions, the detection of Cu 2+ showed good linearity in a wide linear range (0.01 ~ 4 μM). In the analysis of fish samples, the recoveries range from 93% to 96.8% with excellent accuracy, precision and reliability, and show good applicability for the determination of real samples.