Ferrous sulfate hydrate (FeSO₄·7H₂O), urea, ethanol, ammonium molybdate tetrahydrate ((NH₄)₆MoO₂₄·4H₂O), thiourea, bismuth nitrate pentahydrate (Bi(NO₃)₃·5H₂O), glacial acetic acid, glutaraldehyde (50%, GA), ascorbic acid (AA), isopropanol (IPA), p-benzoquinone (BQ), methylene blue (MB), and chitosan (CS) were purchased from Aladdin Reagent Company (Shanghai, China). Fluorine-doped tin oxide (FTO) glass was obtained from South China Xiangcheng Technology Co., Ltd. Oligonucleotides and bovine serum albumin (BSA) were purchased from Sangon Biotech Co., Ltd. (Shanghai, China), and all chemical reagents were analytical grade without further purification.

Aptamer DNA (Apt-DNA): NH 2 − C 12 − CACTACAGAGGTTGCGTCTGTCCCACGTTGTCA TGGG GGGTTGGCCTG

All the electrochemical measurements were carried out on a CHI 760E electrochemical workstation (Shanghai Chenhua Instrument Co., Ltd., China) with a three-electrode system composed of FTO as the working electrode, a platinum electrode as the counter electrode, and a saturate Ag/AgCl electrode as the reference electrode. Electrochemical impedance spectroscopy (EIS) and cyclic voltammetry (CV) were performed in 5 mM K₃Fe(CN)₆/K₄Fe(CN)₆ (0.1 M KCl) as the supporting electrolyte.

Flower-like α-Fe₂O₃ with nanorod petals was prepared as depicted in a previous report with minor modification (Wang et al., 2023c). First, 2.28 g of FeSO₄·7H₂O and 0.6 g of urea were dissolved in 100 mL mixed solution (V_H2O:V_C2H5OH = 4:1) and sonicated for 10 min. Then, the mixed solution was transferred into a 250-mL three-necked flask to reflux at 90°C for 6 h. After precipitation and drying at 60°C for 24 h, the reddish brown FeOOH powder was prepared. Subsequently, the α-Fe₂O₃ nanorods were obtained by the calcination of the prepared FeOOH at 500°C for 3 h in a Laboratory Muffle stove.

α-Fe₂O₃/MoS₂ nanocomposites were successfully prepared via a hydrothermal route. First, 0.1234 g of (NH₄)₆MoO₂₄·4H₂O and 0.2284 g of thiourea were dissolved in 35 mL of distilled water and stirred for 30 min. Then, 0.357 g of α-Fe₂O₃ was added to the above solution under stirring for 30 min. Subsequently, the obtained solution was transferred to a 50-mL Teflon-sealed autoclave and heated to 200°C for 6 h. After being cooled to room temperature, the α-Fe₂O₃/MoS₂ nanocomposites were obtained after being centrifuged and washed three times.

α-Fe₂O₃/MoS₂/Bi₂S₃ nanocomposites were successfully prepared by a hydrothermal process. First, 0.0236 g of thiourea was added in 25 mL distilled water and stirred for 3 min. Then, 0.076 g of Bi(NO₃)₃·5H₂O was added to the above solution and stirred for 20 min. Then, 0.04 g of α-Fe₂O₃/MoS₂ nanocomposites was added and stirred at 180°C for 20 min. After being cooled to room temperature, the α-Fe₂O₃/MoS₂/Bi₂S₃ nanocomposites were obtained after being centrifuged and washed. The product was dried in an oven at 60°C for 24 h for the next experiment.

The PEC aptasensor based on a direct dual Z-scheme α-Fe₂O₃/MoS₂/Bi₂S₃ ternary heterojunction for the ultrasensitive detection of CTCs is shown in Scheme 1. First, 20 μL (2 mg mL⁻¹) of α-Fe₂O₃/MoS₂/Bi₂S₃ nanocomposites were dropped to the surface of FTO, and 20 μL of mixture solution containing chitosan and acetic acid (chitosan/acetic acid = 1%, w/v) was added on the electrode surface of FTO/α-Fe₂O₃/MoS₂/Bi₂S₃. After being dried at 37°C, the FTO/α-Fe₂O₃/MoS₂/Bi₂S₃ electrode was immersed in GA solution (0.2%) and incubated for 30 min. Then, 20 μL of aptamer DNA (5 μM) was dropped onto the electrode and incubated for 40 min at 37°C. Subsequently, 20 μL of BSA (1%) was used to block the nonspecific binding sites, and the capture electrode FTO/α-Fe₂O₃/MoS₂/Bi₂S₃/CS/GA/BSA was constructed. A volume of 20 μL of MCF-7 cell solution with different concentrations was dropped onto the electrode surface and incubated for 120 min at 37°C. Finally, the PEC response of the biosensor was recorded in 10 mL of PBS (0.01 M, pH 7.4) containing ascorbic acid (AA, 0.14 mol L⁻¹) under visible light irradiation using a LED lamp (excitation wavelength, 450 nm; 100 W) with on–off light switching of 10 s.

SEM was used to analyze the morphology of the as-synthesized samples. As shown in Figure 1A, the SEM image of α-Fe₂O₃ displayed a uniform flower-like nanostructure, and the nanorod petal was approximately 3–4 µm in length. The SEM image of α-Fe₂O₃/MoS₂ (Figure 1B) revealed that α-Fe₂O₃ was encapsulated in MoS₂ nanosheets and exhibited ripples, which indicated the formation of the α-Fe₂O₃/MoS₂ heterojunction. Subsequently, Bi₂S₃ grew in the layered MoS₂, and the α-Fe₂O₃/MoS₂/Bi₂S₃ ternary heterojunction exhibited an icicle flower-like structure, as shown in Figure 1C. X-ray diffraction (XRD) patterns were used to characterize the α-Fe₂O₃/MoS₂/Bi₂S₃ ternary heterojunction. Figure 1D shows the characteristic diffraction peaks that correspond to the JCPDS card No. 33–0664 α-Fe₂O₃, respectively. Additionally, three peak representatives (14.13°, 28.47°, and 32.91°) which belonged to the (002), (004), and (100) crystal planes of MoS₂ (JCPDS card No. 75–1539), respectively, proved the formation of MoS₂. Meanwhile, a few prominent peaks of Bi₂S₃ appeared based on JCPDS card No. 17–0320. These illustrated the formation of the α-Fe₂O₃/MoS₂/Bi₂S₃ ternary heterojunction. As expected, the element mapping images (Figure 1E) showed the distribution of O, Bi, Mo, S, and Fe, offering direct evidence of the effective achievement of α-Fe₂O₃/MoS₂/Bi₂S₃ ternary heterojunction. The UV-vis absorption spectra of α-Fe₂O₃, α-Fe₂O₃/Bi₂S₃, and α-Fe₂O₃/MoS₂/Bi₂S₃ were investigated as described in Supplementary Figure S1. Both MoS₂ and Bi₂S₃ presented a broad absorption spectrum across the visible light region. For the α-Fe₂O₃/MoS₂/Bi₂S₃ heterojunction, α-Fe₂O₃ also enhanced its absorption ability in visible light, which would lead to an increase in photocatalytic activity.

The band gap energy (Eg) of α-Fe₂O₃ (2.1 eV), MoS₂ (1.38 eV), and Bi₂S₃ (1.41 eV) was investigated by UV–vis diffuse reflectance spectroscopy, and the flat-band potentials (α-Fe₂O₃, 0.69 eV; MoS₂, −0.4 eV; Bi₂S, 0.59 eV; all vs. SSCE) were derived using Mott–Schottky plots, as shown in Supplementary Figure S2. Their valence bands (VBs) were 0.93 eV, −0.16 eV, and −0.35 eV, respectively, which was obtained based on the following formula: VB = CB + Eg. The trapping experiments of reactive species in this photocatalytic process were carried out. In this work, IPA (radical •OH scavenger) and BQ (radical •O²⁻ scavenger) were employed as quenchers in the degradation experiment of methylene blue (MB), as shown in Supplementary Figure S3. During this photocatalytic process, both BQ and IPA significantly reduced the degradation rate of MB, demonstrating that a larger amount of •OH and •O²⁻ on the surface of α-Fe₂O₃/MoS₂/Bi₂S₃ was involved in the degradation of MB. The standard potential of the OH⁻/•OH pair (+2.40 eV vs. NHE) was lower than the VB position of α-Fe₂O₃ and higher than the VB position of both MoS₂ and Bi₂S₃. We could speculate that only h+ of α-Fe₂O₃ reacted with OH− or H₂O to form •OH. Meanwhile, the standard potential of the O₂/•O²⁻pair (−0.33 eV vs. NHE) was more positive than that of Bi₂S₃ and more negative than the CB of both α-Fe₂O₃ and MoS₂. It was concluded that •O²⁻was more possible to be produced by Bi₂S₃.

Based on these, the transfer pathway of electrons in α-Fe₂O₃/MoS₂/Bi₂S₃ is shown in Figure 2. Under visible irradiation, photo-generated e−/h+ was produced on the CB and VB of α-Fe₂O₃, Bi₂S₃, and MoS₂. The e− in the CB of α-Fe₂O₃ and MoS₂ transferred to the VB of MoS₂ and Bi₂S₃ to recombine with the h+, respectively. This resulted in the accumulation of high-energy e− and h + on the VB of Bi₂S₃ and the CB of α-Fe₂O₃, where they participated in photocurrent production. This double Z-scheme heterojunction promoted the detecting photocurrent intensity in the PEC aptasensor because it effectively inhibited the recombination of electron–hole pairs and absorbed sufficient light.

As shown in Figure 3A, the photocurrent of FTO/α-Fe₂O₃/MoS₂/Bi₂S₃ (curve b) was much larger than that of the naked FTO electrode (curve a) due to the creation of ternary heterojunctions with high light absorption and photoelectric conversion efficiency. Steric hindrance caused a reduction in photocurrents after Apt-DNA (curve c), BSA (curve d), and MCF-7 cell (curve e) were assembled on the photoelectrode of FTO/α-Fe₂O₃/MoS₂/Bi₂S₃. These demonstrated that the PEC aptasensor was successfully constructed. Meanwhile, electrochemical impedance spectroscopy (EIS) was also carried out to validate this process. The diameter of the high-frequency semicircle in the Nyquist plot corresponded to the electron transfer resistance (R_et) of the electrode surface (Luo et al., 2022). As shown in Figure 3B, the R_et value of the naked FTO electrode was small (curve a), and it drastically decreased (curve b) when the α-Fe₂O₃/MoS₂/Bi₂S₃ heterojunction was dropped on the FTO electrode. Subsequently, when the Apt-DNA (curve c), BSA (curve d), and MCF-7 cell (curve e) were continuously assembled on the FTO/α-Fe₂O₃/MoS₂/Bi₂S₃ surface, they led to an increase in R_et because they impeded the diffusion of electrons to the electrode surface, indicating their successful immobilization.

A number of parameters, including the concentration of Apt-DNA, the amount of AA, and the incubation time of the Apt-DNA with the photoelectrode and captured electrode with MCF-7 cells, were optimized. As shown in Figure 4A, the photocurrent response peaked at 0.15 mol L⁻¹, and no obvious change was observed at higher concentrations. As a result, the concentration of AA in all subsequent experiments was 0.15 mol L⁻¹. The impact of Apt-DNA concentration on the PEC response of the biosensor is shown in Figure 4B. The photocurrent decreased as the concentration of Apt-DNA increased up to 5 M, after which there was no obvious change, indicating Apt-DNA saturation. Accordingly, 5 µM of Apt-DNA was used in all subsequent experiments. The immobilization time is also shown in Figure 4C. The photocurrent decreased in the range from 0 to 60 min and then remained constant. It was that the amount of Apt-DNA was saturated after a certain time. Meanwhile, the incubation time of Apt-DNA with the captured electrode was also examined, as shown in Figure 4D. The ideal duration was found to be approximately 120 min. Under optimal conditions, the photocurrent was large and stable, which would be performed for subsequent experiments.

The PEC response decreased with an increase in MCF-7 cell concentrations due to steric hindrance (Figure 5A). The decrease in photocurrent intensity demonstrated a good linear relationship with the logarithm of the MCF-7 cell concentration in the range from 10 to 1×10⁵ cells mL⁻¹ (Figure 5B). The linear regression equation was y = −0.36 lgC cells+2.79 (C cells, cell mL⁻¹) with a correlation coefficient (R ²) of 0.9952 (n = 3) and a low detection limit of 3 cell mL⁻¹ (S/N = 3). Therefore, the PEC aptasensor exhibited an ultrasensitive detection of MCF-7 cells compared with the other biosensors given in Table 1.

To assess the application potential, the prepared PEC aptasensor was used to detect MCF-7 cells in real samples. MCF-7 cells with different concentrations (10, 50, 100, 500, and 1,000 cells mL⁻¹) were spiked into serum samples for the assay. The recoveries of MCF-7 were between 92% and 107.6% with a relative standard deviation (RSD) from 5.7% to 7.8% (Table 2), demonstrating great potential for the detection of CTCs in real samples.