Surface plasmon resonance (SPR) devices are capable of detecting subtle changes in the refractive index (RI) of the external environment with high sensitivity, making them well-suited for a wide range of applications in biomedicine, chemical analysis, high-precision sensing, and environmental monitoring^[1–10]. However, the intrinsic ohmic losses of metals in SPR devices significantly attenuate the resonant signal and broaden the resonance linewidth, ultimately leading to a reduced quality factor (Q-factor)^[11–13]. Recently, quasi-bound states in the continuum (quasi-BICs) have emerged as a promising alternative. Compared with the conventional SPR modes used in traditional SPR sensors, quasi-BICs are a class of modes that emerge at wave vectors slightly deviating from the bound states in the continuum (BICs). BICs, due to their inherent symmetry or destructive interference, do not couple to the external radiation field and, therefore, possess theoretically infinite Q-factors. When the wave vector shifts slightly away from the Γ point, the symmetry or interference conditions are broken, allowing the modes to couple to the external field while still maintaining a very high Q-factor. As a result, utilizing quasi-BICs for sensing can significantly enhance sensor performance because these modes exhibit extremely narrow spectral linewidths, ultra-high Q-factors, and strong field confinement^[14]. In addition, their Q-factors can be flexibly tuned by intentionally breaking the structural symmetry of the device^[14–17]. Consequently, quasi-BICs-based structures have attracted considerable interest in the sensing community, inspiring extensive theoretical investigations and practical implementations^[18–22].

However, previous studies on the application of quasi-BICs in sensing have predominantly focused on the modal characteristics of quasi-BICs, with limited in-depth analysis of how the Q-factor, spectral depth, and Q-factor tunability correlate with overall sensing performance. In fact, both the Q-factor and spectral depth significantly impact key performance metrics such as sensitivity, noise-equivalent limit of detection (LOD), and figure of merit (FOM), all of which govern the sensing accuracy and detection precision^[14,23,24]. For instance, in refractive index sensing, the FOM is determined by the sensitivity of the resonance wavelength to refractive index changes and the spectral linewidth. Meanwhile, the LOD is influenced by the resonance wavelength sensitivity and interrogation-wavelength noise, which includes the noise of the optoelectronic detection system. Additionally, the LOD is affected by both the spectral depth and linewidth^[18,24,25].

In this study, a quasi-BIC is incorporated into an SPR sensor, and its impact on key sensing performance metrics—including sensitivity, LOD, and FOM—is systematically investigated. Theoretical analysis reveals that, compared to conventional SPR sensors, the introduction of a quasi-BIC with an ultra-high Q-factor not only enhances sensitivity but also significantly improves both LOD and FOM. Furthermore, the intrinsic tunability of the Q-factor enables flexible optimization of the device, allowing it to achieve a balance between sensitivity and operational stability across diverse sensing scenarios. Based on this analysis, a high-performance temperature sensor is proposed by integrating a temperature-sensitive material, polydimethylsiloxane (PDMS), into the quasi-BIC structure. Simulation results demonstrate that the sensor achieves a high temperature sensitivity of −0.33nm/°C, while maintaining low noise and high temperature resolution. This theoretical framework provides valuable insights for the practical deployment of quasi-BICs in advanced sensing applications. The proposed sensor, featuring high sensitivity, low noise, and tunable Q-factor within a structurally simple design, offers a promising pathway toward next-generation photonic temperature sensing systems tailored for precision measurement in dynamic environments.

The schematic of the proposed structure is illustrated in Fig. 1(a). It consists of a silica glass substrate on which a one-dimensional silver grating is fabricated. This grating is encapsulated by a layer of PDMS, a temperature-sensitive material, with air serving as the external medium. The use of PDMS enhances the sensor’s temperature sensitivity due to its high thermo-optic coefficient (TOC)^[26,27]. The incident light is polarized in the XY plane, with an incidence angle denoted by θ. Figure 1(b) presents the cross-sectional view of the device in the XY plane. The silver grating has a period of P=720nm. The total thickness of the grating—comprising both the ridge height and the underlying silver film—is t=130nm. The ridge widths are specified as a1=180nm and a2=270nm, and the thickness of the back silver film is h=100nm. The PDMS layer has a thickness of d=500nm.

BICs are special eigenstates that reside within the continuum spectrum yet remain completely decoupled from radiative modes. In theory, BICs exhibit an infinite Q-factor due to the absence of energy leakage into the surrounding continuum^[21]. However, in practical implementations, several factors—such as the finite number of structural units, intrinsic material losses, and fabrication imperfections—inevitably lead to coupling with the continuum. As a result, the ideal BIC transitions into a quasi-BIC, characterized by a finite but still potentially high Q-factor^[28].

The subsequent analysis focuses on the device’s reflectance spectra, as quasi-BICs exhibit angularly dependent spectral features—characteristic of their behavior in momentum (k-) space, as illustrated in Fig. 2. Figure 2(a) presents simulated reflectance spectra at incident angles of 0° and 2°. A sharp Fano resonance, indicated by the green arrow, abruptly appears at θ=2°. This phenomenon constitutes a definitive signature of quasi-BIC operation, arising from symmetry-broken coupling between guided modes and radiative continuum states^[21,29,30]. At θ=0° and θ=2°, a conventional SPR mode appears in the reflection spectrum (light red dotted box). The simulated reflection spectrum as a function of the angle of incidence (band diagram) is shown in Fig. 2(b). At normal incidence, vanishing linewidths are observed at the green [corresponding to the green circle in Fig. 2(a)] and purple circles (refer to the Supplementary Material), indicating the presence of symmetrically protected BIC^[21]. The quasi-BIC mode indicated by the purple circle at low resonant wavelengths is an optical mode. Therefore, we chose the quasi-BIC mode represented by the green circle, which is a plasmonic mode with more strongly confined electric fields. Additionally, the vanishing linewidth at the white circle signifies a BIC located at an off-Γ point, formed via interference between different radiation channels, in line with the Friedrich–Wintgen mechanism^[21].

Figure 2(c) presents the reflectance spectrum at an incidence angle of θ=2°, corresponding to the conventional SPR mode. The inset shows the electric field distribution, which reveals a broad spectral linewidth and field localization primarily at the silver surface. In contrast, the detailed spectrum of the quasi-BIC mode is shown in Fig. 2(d). Compared to the conventional SPR mode, the quasi-BIC exhibits a significantly narrower linewidth, corresponding to a high Q-factor of up to 1075. Additionally, the spectral dip of the quasi-BIC approaches 100%, indicating strong suppression of background reflection—an attribute that contributes to reduced noise, a lower LOD, and an improved FOM for the device. The inset of Fig. 2(d) illustrates the corresponding electric field distribution, demonstrating substantially enhanced field confinement at the silver surface compared to the conventional SPR mode.

The previous section provided a detailed analysis of the quasi-BIC characteristics. In the following section, we focus on the critical correlation between the Q-factor of the quasi-BIC and key performance metrics of the SPR sensor, including sensitivity, LOD, and FOM. Both qualitative and quantitative relationships between the Q-factor and these sensing parameters are examined. Finally, a comparative analysis is performed between the quasi-BIC and conventional SPR modes to highlight the differences in sensing performance, particularly in terms of sensitivity, LOD, and FOM.

For refractive index sensing applications, such as biomolecule detection and temperature monitoring, the detection process typically involves tracking the centroid wavelength shifts of the device’s resonance spectra^[24,25]. As shown in Fig. 3(a), these centroid wavelength shifts serve as the sensing signal. The spectra exhibit a depth of approximately 50%, with a full width at half-maximum (FWHM) of 4.25 nm, corresponding to a Q-factor of approximately 236. The light brown and light blue curves represent SPR spectra measured at different sample concentrations or temperatures, corresponding to variations in refractive index. Additionally, the threshold used in the centroid wavelength algorithm is denoted as η, with only spectral regions below this threshold being considered for analysis^[22]. Figure 3(b) presents the schematic of the quasi-BIC employed for sensing. Compared to the conventional SPR mode, the quasi-BIC demonstrates a significantly higher Q-factor (∼1075), a narrower FWHM (∼1nm), and a nearly complete spectral depth (∼100%).

The wavelength interrogation noise in this scheme satisfies Eq. (1): λc−λ−=∑i(λi−λ−)Ii∑iIi,(1)where λc is the experimentally measured centroid wavelength, λ¯ is the exact centroid wavelength, and i is the numbering of each pixel of the spectrometer. λi is the wavelength of the pixel i, and Ii is the photon intensity of the pixel i. Consider that the measurement noise (Nλ) of λc mainly comes from the detected photon intensity noise, which is mainly associated with shot noise^[25]. According to Eq. (1), the primary noise source will be the detection noise at locations far from the resonant wavelength. In other words, a higher spectrum depth and low background noise at non-resonant wavelengths could reduce the interrogation noise of the device^[24,25]. Additionally, the Sλ, LOD, and FOM of the device satisfy Eqs. (2)–(4): Sλ=ΔλΔn,(2)LOD=3×NλSλ,(3)FOM=SλΓλ,(4)where Nλ, Sλ, and Γλ are the interrogation noise, the wavelength sensitivity of the device, and the FWHM of the reflection spectrum, respectively^[18,31]. Moreover, for a given threshold (η), quasi-BICs exhibit a considerably higher Q-factor and a more profound spectrum depth, and hence, the device achieves a lower LOD. Furthermore, according to Eq. (3), the FOM of the device is expected to undergo significant optimization due to the quasi-BICs with high Q-factors (corresponding to a low FWHM). Consequently, the device is anticipated to achieve a high level of biomolecule detection and excellent temperature-sensing accuracy^[24].

A comparison of the Sλ, Q-factor, and FOM parameters of the devices in quasi-BIC and SPR modes is presented in Table 1. The findings indicate that quasi-BIC exhibits a substantial superiority over the SPR mode in terms of Q-factor and FOM. The precise LOD values are not available due to the lack of specific noise experimental results. However, according to the wavelength interrogation noise Eq. (1), it is concluded that the noise of quasi-BIC is significantly lower than that of the SPR mode. Furthermore, given that the wavelength sensitivity (Sλ) of quasi-BIC is slightly higher than that of the SPR mode, the LOD parameter of quasi-BIC is superior to that of the SPR mode.

In addition to noise reduction, the high Q-factor can enhance the device’s sensitivity for optical intensity-modulated sensing applications. Figure 4(a) shows the principle of optical intensity-modulated sensing (e.g.,  acoustic pressure sensing) using the SPR mode. The results demonstrate that, under the effect of the external tiny sinusoidal acoustic pressure (corresponding interrogated wavelength offset Δλ), the sensing response (ΔR) is the corresponding reflection power changes^[32]. In practical sensing systems, converting optical signals into electrical signals by optical receivers facilitates signal acquisition and subsequent processing^[33–36]. As illustrated in Fig. 4(b), the sensing response of quasi-BICs for detection is demonstrated. According to the results, the sensing response of quasi-BICs is higher than that of the SPR mode. For the acoustic pressure sensing application, the intensity sensitivity, Sac, of the device satisfies Eq. (5), Sac=dRrdλ×dλdn×dndp,(5)where Rr is the reflectivity, which is directly related to the reflected optical power of the device, and λ is the wavelength. The n and p are the refractive indices of the acoustic-sensitive material and the acoustic pressure acting on the device, respectively^[37]. After the theoretical derivation, it is concluded that the sensitivity of the device satisfies Eq. (6), Sac=dRrdλ×dλdn×(n0−1)×Y−1,(6)where n0 and Y are the refractive index and Young’s modulus of the acoustic pressure-sensitive material (at zero acoustic pressure), respectively; therefore, one of the key factors determining the sensitivity of the device is the slope of the interrogated wavelength on the reflectivity curve k (k=dRr/dλ), and k is the light intensity sensitivity in mW/nm. Combined with Eq. (6), the acoustic pressure sensitivity exhibits a clear linear relationship with the k under no change in other parameters. Additionally, the Lorentz formula is used to derive the acoustic pressure sensitivity versus the Q-factor (for the Lorentz curve fitting of the reflectance curve, please refer to the Supplementary Material).

Assuming that the reflectivity curve Rr(λ) is expressed as Eq. (7) in terms of the Lorentz curve, Rr(λ)=1−A1+[2(λ−λr)Γλ]2(7)

In Eq. (7), A is the depth of the spectra. λr and Γλ are the resonance wavelength and the FWHM, respectively. The first-order derivatives of the reflectivity curve are used to obtain Eq. (8), |dRr(λ)dλ|=8A(λ−λr)Γλ2{1+[2(λ−λr)Γλ]2}2(8)

Consider that the Q-factor is defined as Eq. (9), Q=λrΓλ.(9)

Assuming that the interrogated wavelength of the device is λ0=λr±mΓλ, the slope k is obtained by substituting Eq. (9) into Eq. (8), |k|=8m(1+4m2)2×A·Qλr.(10)

Substituting Eq. (10) into Eq. (6) for the sensitivity of the device, the sensitivity Sac is obtained to satisfy Eq. (11), Sac=8m(1+4m2)2×A·Qλr×dλdn×(n0−1)×Y−1.(11)

The analysis indicates that the sensitivity Sac is proportional to the product of the depth of the spectra and the Q-factor under the condition that all other variables remain constant. Consequently, quasi-BICs, characterized by a high Q-factor and depth of the spectrum, exhibit a substantial enhancement in the sensitivity of the device or the sensing system in light-intensity detection applications, such as acoustic pressure sensing.

Figure 4(c) illustrates the Q-factor of the quasi-BIC as a function of the incidence angle of the incoming light. The results show that the Q-factor decreases as the incident angle increases from 1° to 6°. In other words, the Q-factor of the quasi-BIC can be actively tuned by varying the angle of incidence, which effectively breaks the structural symmetry^[18,38]. Consequently, the slope k of the interrogated resonance wavelength changes accordingly, leading to modulation of the device’s intensity sensitivity, Sac, of the device. This tunability not only enables a significant enhancement of intensity sensitivity but also allows the device to maintain an optimal balance between sensitivity and stability under different operational conditions^[39]. Furthermore, the resonance wavelength of the quasi-BIC can be tailored to suit sensing systems operating at different wavelengths by adjusting the silver grating periods a1 and a2 (see the Supplementary Material).

The comparison of the Q-factor and intensity sensitivity parameters of the device in quasi-BIC and SPR modes is demonstrated in Table 2. The findings indicate that the quasi-BIC exhibits a substantial advantage over the SPR mode with respect to the Q-factor and intensity sensitivity. At this stage, the quasi-BIC serves to enhance the intensity sensitivity of the device.

A high-performance temperature sensor based on the quasi-BIC is developed by integrating the temperature-sensitive material PDMS. Figure 5(a) presents the simulated reflectance spectra of the quasi-BIC mode at an incidence angle of θ=2°, showing the relationship between the centroid wavelength and the refractive index of PDMS. The results demonstrate that the centroid wavelength undergoes a red shift as the PDMS refractive index increases. Although the SPR spectrum exhibits a similar trend in centroid wavelength shift, the quasi-BIC—with its higher Q-factor and greater spectral depth—provides reduced noise levels, thereby enhancing sensing performance^[24,25]. It is worth noting that other spectral features are not considered for refractive index sensing in this study; detailed mode analyses of these features can be found in Ref. [21].

Figure 5(b) depicts the relationship between the centroid wavelength of the quasi-BIC and the refractive index of PDMS. The blue data points represent the simulation results, while the red curve corresponds to a linear fit. The fitting reveals that the device exhibits a refractive index sensitivity of 725 nm/RIU. In Fig. 5(c), the left panel shows the electric field distribution of the quasi-BIC and SPR modes at the resonance wavelength on the surface of the metallic silver. The right panel displays the field distribution at 1 nm detuned from the resonance wavelength. It is clear that the electric field intensity of the quasi-BIC mode is significantly stronger than the SPR mode, confirming that the quasi-BIC mode possesses a narrow FWHM, consistent with a high Q-factor from an electromagnetic field perspective^[21,40].

By incorporating the TOC of PDMS (−4.5×10−4/°C) into the model, the finite element method simulations of the quasi-BIC spectra were performed, as shown in Fig. 5(d). The centroid wavelength of the quasi-BIC exhibits a gradual blue shift with increasing temperature^[27]. It is important to note that the TOCs of silica glass and metallic silver are approximately 2 orders of magnitude smaller than that of PDMS; therefore, their influence on the temperature sensitivity is negligible and not considered in this analysis. Figure 5(e) summarizes the simulated temperature sensitivity of the device, where the blue data points represent the simulation results, and the red line corresponds to the linear fit. The fitting yields a temperature sensitivity of −0.33nm/°C, with excellent linearity.

A comparison of the Q-factor and temperature sensitivity of the devices in quasi-BIC and SPR modes is shown in Table 3. The findings indicate that, with regard to temperature sensitivity, the discrepancy between quasi-BIC and SPR modes is not readily apparent, primarily because the wavelength sensitivities of the two modes are more similar to each other. Consequently, for the same temperature-sensitive material PDMS, the discrepancy in temperature sensitivity between the two modes is not readily apparent.

In addition, according to the principle of refractive index sensing, the temperature sensitivity of the device satisfies Eq. (12): ST=dλdn×dndT(12)derived from the concept of the TOC, TOC=dndT,(13)ST=dλdn×TOC.(14)

Substituting the relevant data (refractive index sensitivity dλ/dn=725nm/RIU) into Eq. (13) yields the temperature sensitivity of the device in the quasi-BIC as −0.33nm/°C, which corresponds to the simulation results. Notably, the temperature sensitivity of the proposed device significantly surpasses that of many previously reported quasi-BIC-based temperature sensors^{[15,18,41,42]}. This improvement can be attributed to the combination of the device’s high refractive index sensitivity and the incorporation of a temperature-sensitive material with a large TOC^[27]. Moreover, the device exhibits excellent LOD and FOM, which translates to enhanced temperature detection accuracy and resolution. These advantages stem from the ultra-high Q-factor and nearly complete spectral depth characteristic of the quasi-BIC mode.

A comparative analysis of the performance of this study with the latest reported performance of the temperature sensors is presented in Table 4. The findings indicate that the device exhibits no discernible superiority in terms of Q-factor and FOM, a phenomenon primarily attributable to ohmic losses, a consequence of its metallic composition. Nonetheless, the devices exhibit notable advantages in terms of wavelength and temperature sensitivity. Additionally, the device demonstrated notable temperature-sensing stability across a range of incidence angles (1–6), indicating its exceptional angular robustness. The sensor’s simple structural design, coupled with its high sensitivity, low noise, and tunable Q-factor, highlights its strong potential for applications in next-generation high-performance photonic temperature sensing systems.

In summary, quasi-BICs possess several advantageous properties, including an exceptionally high Q-factor, substantial spectral depth, and significant tunability of the Q-factor. As a result, SPR sensors based on quasi-BICs have attracted considerable attention for high-performance photonic sensing applications. Although extensive research has focused on mode analysis of quasi-BICs, less emphasis has been placed on understanding the intrinsic sensor parameters such as intensity sensitivity, LOD, and FOM. In this study, we introduce a quasi-BIC into an SPR sensing device and systematically analyze how the high Q-factor influences its sensing performance, including intensity sensitivity, LOD, and FOM. The results demonstrate that the quasi-BIC’s high Q-factor enhances intensity sensitivity, reduces noise, and optimizes both LOD and FOM. Additionally, the tunability of the Q-factor enables the device to achieve a desirable balance between sensitivity and stability across different applications. By integrating the temperature-sensitive material PDMS, we further developed a temperature sensor achieving a sensitivity of −0.33nm/°C, alongside excellent theoretical LOD and FOM. This work establishes a fundamental relationship between the SPR device performance and the quasi-BIC’s high Q-factor, offering a promising approach for the flexible design of high-performance photonic temperature sensors based on the quasi-BIC.