Lung cancer is one of the most common and serious types of cancer that ranks high in both incidence and mortality rates among all types of cancers. The prevention, early detection, and timely intervention of lung cancer are crucial in controlling its incidence and mortality rates^[1,2]. However, existing detection methods for lung cancer face issues, such as low sensitivity, poor specificity, high costs, and invasiveness, which limit their application in early lung cancer detection. There is a pressing need to develop new, efficient, and non-invasive detection technologies.

Terahertz (THz) metasurfaces can significantly enhance the local electromagnetic field intensity, improving the ability to detect weak biological signals^[3–6]. They combine high sensitivity, high specificity, and rapid detection with the non-destructive and penetrating capabilities of THz electromagnetic waves for biological tissues^[7,8]. These properties make THz metasurfaces suited for detecting and monitoring living cells and tissues, offering significant advancements over traditional methods and becoming a cutting-edge technology in the field of biosensing. With the development of micro-nano optoelectronic devices, novel physical mechanisms are constantly emerging. THz metasurfaces supported by bound states in the continuum (BICs) have shown tremendous potential for highly sensitive biological detection^[9–12]. BICs are the localized states that exist in the continuum, characterized by zero radiation loss and infinite quality factors (Q-factors), which can achieve a significant enhancement of the local electromagnetic field^[13–18]. By precisely adjusting the structural parameters, the BIC can be transformed into observable quasi-BICs (QBICs) with tunable narrow line widths and high Q-factors, enabling high-sensitivity detection of biomolecules. At present, QBIC-based metasurfaces are widely promoted in the field of biosensing for detecting substances^[19–23], such as amino acids^[19], c-reactive protein^[20], and homocysteine (Hcy) molecules^[23]. However, most measurements require the samples to be dried to eliminate the impact of water molecules on the terahertz waves. For cancer cells, living cell detection aids in screening and diagnosing malignancies, providing more comprehensive diagnostic information. It is crucial to explore an effective, rapid, real-time, and highly sensitive live-cell detection scheme.

Here, we design a QBIC-based THz biosensor and analyze its sensing capability for different concentrations of lung cancer cells (as shown in Fig. 1). By breaking the structural symmetry, the BICs can transition into QBICs. Asymmetric parameters are introduced to adjust the line widths and Q-factors of QBICs. The calculated refractive index sensitivity is 354 GHz/RIU, which indicates that the designed biosensor exhibits excellent refractive index sensing performance. The micro-channel liquid sample pool is combined with the designed QBIC biosensor to establish a micro-volume detection system. The resonant frequency and intensity of the measured spectrum changed obviously with the increase in cell concentration. Our work provides a novel approach for the detection of living lung cancer cells, offering promising applications in apoptosis detection and providing more effective solutions for the early detection and intervention of lung cancer.

The schematic of the unit cell is presented in Fig. 2(a). The materials of the metal structure and substrate are gold and quartz, respectively, with thicknesses of 0.2 and 500 µm. The periods in the x-and y-directions are px=py=60μm. The outer radius r1 of the split-ring resonator is 25 µm, the inner radius r2 is 13 µm, and the fixed gap g is 8 µm. g1 is an adjustable parameter that determines the bandwidth of the spectra. Figure 2(b) depicts the simulated spectra as g1 changes from 8 to 26 µm. When g1=8μm, the spectral linewidth disappears at 1.4 THz, indicating the presence of a BIC with an infinite Q-factor. With the increase of g1, the spectral line width broadens, indicating that the radiation loss gradually increases. At this time, the surface current in the xoy plane forms a ring-like distribution, generating a magnetic field along the z-axis, as shown in Fig. 2(c). The increase in g1 leads to an enhancement in magnetic field leakage. Figure 2(d) plots the simulated transmission spectrum of the structure with g1=24μm, where a fano resonance is observed at 1.66 THz. The calculated scattering power of the electromagnetic multipole is presented in Fig. 2(e), including the electric dipole (ED), the magnetic dipole (MD), the toroidal dipole (TD), the electric quadrupole (EQ), and the magnetic quadrupole (MQ). Near 1.66 THz, the contribution of the MD is the largest, which indicates that the generation of QBIC primarily originates from the MD, consistent with the analysis in Fig. 2(c).

The moments and scattered power of each multipole are expressed as^[24]$$\begin{gathered} ED:P=1iw∫jd3r, \\ MD:M=12c∫(r×j)d3r, \\ TD:T=110c∫[(r·j)r−2r2j]d3r, \\ EQ:Qα,β(e)=1i2w∫[rαjβ+rβjα−23(r·j)δα,β]d3r, \\ MQ:Qα,β(m)=13c∫[(r×j)αrβ+(r×j)βrα]d3r, \\ I=2w43c3|p|2+2w43c3|M|2+4w53c3|p·T|+2w63c5|T|2 \\ +w65c5∑|Qα,β(e)|2+w620c5∑|Qα,β(m)|2+o(1c5), \end{gathered}$$where j represents the surface current density and c is the speed of light. The total radiated power I in the far field is determined by the sum of radiation powers from each multipole moment.

To further verify the existence of BIC, we analyze the optical band and radiative Q-factor of the structure. When the structure maintains in-plane symmetry (g=g1=8μm), the calculated optical bands and Q-factors of the structure are shown in Figs. 3(a) and 3(b). At the Γ point, a TE band with an infinite Q-factor is observed at 1.4 THz. Away from the Γ point, the Q-factor decreases significantly. Figure 3(c) illustrates the variation of the resonance frequency and Q-factor with different g1. As g1 increases from 8 to 26 µm, the coupling between QBICs and the radiative continuum results in a significant decrease in the Q-factor, accompanied by a shift of the resonance frequency toward higher frequencies. The asymmetry parameter is defined as δ=(g1−g)/g. The Q-factor of the QBIC is inversely proportional to the square of δ, as shown in Fig. 3(d). These characteristics indicate that the designed structure is a typical symmetry-protected BIC, which allows for the tuning of the resonance frequency and Q-factor of QBIC by changing the asymmetry parameter.

A series of sensors with different disturbance parameters are fabricated for experimental verification. The total size of the fabricated sensor is 1cm×1cm, consisting of 100×100 arrays. The simulated and measured spectra are obtained under y-polarized incidence. Figures 4(a) and 4(b) indicate that the measured results agree well with the simulated results. As g1 increases, the resonance frequency gradually shifts to higher frequencies, and the linewidth broadens. The difference in the spectral resonance linewidth between the simulated and the measured spectra is due to the fabrication error and the inherent loss of the metal material. When g1=24μm, the Q-factors calculated by simulation and experiment are 40 and 16, respectively. The lower experimental Q-factor compared to the simulation value is attributed to the ohmic and scattering losses of the metal structures, sample fabrication errors, and the limited resolution of the THZ-TDS system. For symmetry-protected BIC, the Q-factor of QBIC can be tuned by changing the asymmetry parameter to match the requirements of the test system resolution and practical application.

Next, the refractive index sensing characteristics of the designed biosensor (g1=24μm) are analyzed. In the simulation, the analyte with a thickness of 20 µm is set on the surface of the structure, which corresponds to the actual size of A549 lung cancer cells^[25]. The simulated spectra of the biosensor with the refractive indices n varied from 1 to 2 are plotted in Fig. 4(c). An increase in the refractive index results in a noticeable red shift of the transmission spectrum. The calculated refractive index sensitivity is 354 GHz/RIU [Fig. 4(d)].

The resonance intensity and Q-factor of the sensor are tuned by asymmetric parameters, while the sensing sensitivity is largely determined by the structural period. We analyze the refractive index sensitivity of the same structure with different periods. To eliminate the effects of other structural parameters, such as substrate and metal structure thickness, split-ring radius, and gap size, the scaling factor parameter is defined. Keeping the substrate and metal structure thickness constant, the other structure parameters are proportionally scaled in the x-and y-directions based on the parameter settings in Fig. 2(a). The impact of the structural period on sensor performance is systematically studied. The results, shown in Fig. 4(e), highlight the sensitivity changes for periods of 30, 60, and 90 µm, corresponding to scaling factors of 0.5, 1, and 1.5 respectively. As the scaling factor c decreases from 1.5 to 0.5, the sensitivity of the sensor increases from 232 to 740 GHz/RIU. This may be attributed to the stronger localized electric field generated by the smaller periodic structure, which significantly amplifies the response to environmental signals. The variation in the radiative Q-factor with asymmetric parameters, as shown in Fig. 4(f), follows the formula Q∝mδ−2^[26–28], where m is a constant determined by the geometric parameters of the structure. This trend is consistent across different scaling factors with c=0.5, 1, and 1.5. It is noteworthy that as the period decreases, the radiative Q-factor of the structure decays more rapidly with increasing asymmetric parameters. This implies that in practical fabrication, smaller manufacturing tolerances are required to achieve the target Q-factor. Balancing the need for high sensitivity with manufacturing feasibility and cost, a sensor with a period of 60 µm was chosen for biomolecular sensing. This decision ensures that the sensor maintains sufficient sensitivity while being practical for manufacturing and experimental implementation.

The transmitted spectra of the sensor are measured utilizing a THz-time-domain spectroscopy (THz-TDS) test system. Figure 5(a) displays the schematic diagram of the testing system. The microscopic images of the fabricated sensor are presented in Fig. 5(b), where g1=24μm. To address the challenge of strong terahertz signal attenuation by water in liquid samples, a specially designed liquid sample pool was developed for detecting various concentrations of lung cancer cells. Figure 5(c) shows the schematic diagram of the designed sample cell, which allows for precise control of the liquid sample volume on the sensor surface, providing great convenience for cell concentration detection.

A series of lung cancer cell samples of different concentrations are prepared for measurement. Figure 6(a) presents the microscopic images of sensors covered with different concentrations of living lung cancer cells, where the concentrations are 40×105 and 5×105cells/ml, respectively. The measured spectra of the sensor with different concentrations are shown in Fig. 6(b). The increase in the concentration of lung cancer cells affects the refractive index of the solution, leading to a red shift in the transmitted spectrum. Figure 6(c) presents a magnified view of the resonance dips, which provides a more intuitive and detailed observation of the changes in resonance frequency and intensity. When the concentration of the lung cancer cells increases from 2.5×105 to 40×105cells/ml, the resonance frequency shifts from 1.364 to 1.32 THz, and the resonance intensity increases from 0.223 to 0.327. Moreover, as the cell concentration decreases, the water content in the solution increases, enhancing the absorption of the terahertz waves, which leads to a reduction in the resonance depth. According to the measured results, the variation curves of the resonant frequency and amplitude with cell concentration were fitted according to linear and logarithmic relations respectively in Fig. 6(d). The linear fitting curve for the frequency follows f=1.359–0.001C, where C represents the cell concentration, and the coefficient of determination R2=0.95. The measured results are consistent with the simulated results. The logarithmic fitting curve for the amplitude follows A=0.24−0.02[ln(C)−2.06], and the coefficient of determination R2=0.99.

After each measurement, the sensor was cleaned in ultrapure water to carefully remove residual cells from the sensor surface and then dried. Before the next measurement, the transmitted spectrum of the cleaned sensor was measured to ensure that the cleaning process did not affect its optical response. As shown in Fig. 6(e), the transmitted spectra of the bare sensor and the cleaned sensor after measuring different cell concentrations are highly consistent, demonstrating the reusability of the device. This design facilitates real-time, rapid detection, and analysis of living lung cancer cells, providing a feasible method for detecting cancer cell apoptosis.

In conclusion, we design a QBIC-based THz biosensor for lung cancer cell sensing. By introducing asymmetric parameters, the in-plane symmetry of the symmetry-protected BIC is broken and QBIC with a tunable Q-factor is obtained. The sensing ability of the designed sensor on living lung cancer cells is investigated. The simulation and experimental results indicate that the refractive index sensitivity of the designed sensor is 354 GHz/RIU, which can realize the direct detection of different concentrations of lung cancer cells. Our design provides new insights for the application of QBIC biosensors in the biomedical field and promotes their practical use in the early diagnosis of lung cancer.

Numerical Simulation: The simulation of the transmitted spectrum and sensing performance of the designed QBIC sensor was carried out utilizing the CST Microwave Studio with the time-domain solver. The periodic boundary conditions were set in the x- and y-directions and open boundary conditions were set in the z-direction. The permittivity of the quartz was set to 3.79, and the electrical conductivity of gold was 3.56×107Sm−1. The y-polarized waves were incident vertically along the z-direction. The analysis of the optical band and the radiative Q-factor were completed using the COMSOL Multiphysics software. The unit cell was set to Floquet periodic boundary conditions, and perfect matching layers were utilized for the input and output ports. The metal structures were set as the perfect electric conductor (PEC).

Sensor Fabrication: The designed sensor was prepared using traditional photolithography techniques. First, gold with a thickness of 20 µm was deposited on the cleaned quartz substrate by magnetron sputtering. The photoresist was uniformly spin-coated onto the metal layer. After the soft bake, the mask with a structured pattern was exposed to ultraviolet light for development. Then, the pattern was etched to obtain the structural array. Finally, the lift-off process was completed and then the sensor was cleaned and dried.

Measurement of Transmission Spectrum: The transmitted spectrum of the sensor was measured using the THz-TDS system. The ultrashort pulse beam emitted by a femtosecond laser was used as the light source. The THz wave emitted from the transmitter was illuminated to the sensor through Polarizer 1 and Lens 1. The receiver collected the THz signal that had passed through Lens 2 and Polarizer 2 to obtain spectral information. To mitigate the adverse absorption of the THz signal by water vapor, the measurement was performed in an enclosed space filled with nitrogen gas. During the measurement, the environmental temperature was kept at 18°C and the relative humidity was kept at less than 5% RH.

Cell Culture and Sample Preparation: The culture process of the A549 lung cancer cells was as follows:

1. 1.Discard the culture supernate and rinse the cells with 2 ml phosphate buffered brine (PBS).
2. 2.Add 2 ml trypsin EDTA into the culture dish, and put it in an incubator to digest for 3 min at 37°C.
3. 3.Centrifuge the cytochylema at 1200 RPM for 3 min and discard the culture supernate.
4. 4.Add 1 ml PBS and blow continuously to form a uniform cell solution.

The cell solution with 10 µl was placed on a blood cell counting plate and counted under a microscope. Samples with different concentration gradients were obtained by diluting them with PBS buffer (40×105, 20×105, 10×105, 5×105, and 2.5×105cells/ml). Before measuring, the sensor surface was cleaned with PBS. The sensor was placed into the sample cell, and a pipette was used to draw an appropriate amount of cell solution. The cell solution was slowly injected into the sample cell, ensuring it evenly covered the sensor surface.