Numerical and Experimental Demonstration of a Silicon Nitride-Based Ring Resonator Structure for Refractive Index Sensing

Abstract

In optical communication and sensing, silicon nitride (SiN) photonics plays a crucial role. By adeptly guiding and manipulating light on a silicon-based platform, it facilitates the creation of compact and highly efficient photonic devices. This, in turn, propels advancements in high-speed communication systems and enhances the sensitivity of optical sensors. This study presents a comprehensive exploration wherein we both numerically and experimentally display the efficacy of a SiN-based ring resonator designed for refractive index sensing applications. The device’s sensitivity, numerically estimated at approximately 110 nm/RIU, closely aligns with the experimental value of around 112.5 nm/RIU. The RR sensor’s Q factor and limit of detection (LOD) are 1.7154 × 10⁴ and 7.99 × 10⁻⁴ RIU, respectively. These congruent results underscore the reliability of the two-dimensional finite element method (2D-FEM) as a valuable tool for accurately predicting and assessing the device’s performance before fabrication.

1. Introduction

Silicon nitride (Si₃N₄) [1], indium phosphide (InP) [2], and silicon-on-insulator (SOI) [3] are pivotal materials in the realm of integrated photonic sensors, each presenting distinct benefits and limitations [4]. Silicon nitride (Si₃N₄) stands as a pivotal cornerstone in the landscape of photonic devices, presenting a highly adaptable platform with an expansive array of applications [5,6]. The efficacy of Si₃N₄ photonic devices is rooted in the distinctive optical properties inherent to this material, encompassing low optical losses, a broad transparency range, high nonlinear coefficients, and seamless compatibility with complementary metal oxide semiconductor (CMOS) technology [7,8]. On the other hand, InP is a highly desirable material due to its direct bandgap, facilitating efficient light emission and amplification, which is essential for active photonic devices like lasers and amplifiers, thereby making it indispensable in telecommunications and high-speed data transmission. Si, while also offering low-cost and well-established fabrication techniques, is typically used for passive components due to its indirect bandgap. Nevertheless, SOI photonics has surged in popularity because it allows for the integration of electronic and photonic components into a single chip, utilizing mature CMOS technology [9]. In summary, Si₃N₄ is superior for low-loss, broad-spectrum applications, InP is unparalleled for active photonic devices, and SOI provides a flexible and cost-effective solution for integrating electronics with photonics. The selection of the material hinges on the specific needs of the photonic sensor application, such as the required wavelength range, integration considerations, and the focus on active or passive components [10].

Ring resonator (RR) sensors offer superior performance compared to other photonic sensing devices due to their exceptional sensitivity and compact design [11,12]. Their high quality (Q) factor allows for enhanced interaction between light and the analyte, leading to highly precise detection of minute variations in the refractive index or absorption. Unlike larger devices such as Mach–Zehnder interferometers (MZIs) [13,14], RRs are space-efficient, enabling integration into miniaturized, on-chip platforms ideal for portable and lab-on-a-chip applications. Additionally, their design supports high multiplexing capabilities, allowing for the simultaneous detection of multiple analytes, and is compatible with cost-effective silicon photonics fabrication processes. These advantages make RR sensors a versatile and powerful choice for various high-precision sensing applications. RR sensors have been fabricated on several material platforms, including SOI, to leverage mature silicon photonics technology [15], and Si₃N₄, due to its low optical losses and broad transparency range [16]. III-V semiconductors like InP [17] and gallium arsenide (GaAs) [18] are used due to their excellent optoelectronic properties, enabling integration with active components. Polymers [19] and hybrid plasmonic materials [20] are explored for flexible, cost-effective designs, particularly in biosensing and chemical detection. Additionally, glass and quartz substrates are utilized due to their high transparency and low optical losses, supporting high-quality sensing performance across various applications.

Si₃N₄ waveguides excel in facilitating efficient light propagation, ensuring the seamless routing of optical signals with minimal losses. RRs constructed from Si₃N₄ showcase robust light confinement, fostering heightened light–matter interactions and finding utility in diverse applications such as sensors and filters [21,22]. Furthermore, Si₃N₄ modulators capitalize on the material’s potent electro-optic effect, offering a dynamic means to actively control the intensity of light signals [23]. The synergy of Si₃N₄ with standard semiconductor fabrication processes positions it as a compelling choice for integrated photonic circuits, driving advancements in communication, sensing, and quantum information processing [24]. The versatility and exceptional performance of Si₃N₄ in the realm of photonic devices underscore its pivotal role in the continuous evolution of integrated optics [16].

The RR structure holds immense significance in the realm of sensing technologies, offering a novel and highly sensitive approach to detecting various physical and chemical parameters [4]. Utilizing the principles of light confinement within a circular waveguide, these RRs enable the precise interrogation of changes in their immediate environment [25]. This technology has been proven instrumental in applications such as biosensing, environmental monitoring, and telecommunications. Its high sensitivity allows for the detection of minute changes in the refractive index, making it particularly valuable in medical diagnostics and the study of biological interactions at the molecular level [26]. Moreover, the compact nature of photonic RRs enhances their integration into micro- and nanoscale devices, paving the way for advancements in miniaturized sensing platforms [27]. As researchers continue to explore and refine this technology, the RR sensor stands as a beacon of innovation, promising breakthroughs in fields where precision and sensitivity are paramount [28].

Refractive index sensing assumes a crucial role in advancing label-free biosensing methodologies, offering a direct and highly sensitive approach to detecting molecular interactions. In the context of label-free biosensing, alterations in the refractive index become a key parameter for monitoring changes in the optical characteristics of a sensing surface resulting from biomolecular binding events [29]. As biomolecules bind to the sensor surface, they induce variations in the local refractive index, leading to observable shifts in the resonance angle or interference patterns [30]. These shifts exhibit high sensitivity and are proportionate to the mass and concentration of the bound molecules. The real-time measurement of these changes allows us to gather valuable insights into the kinetics, affinity, and concentration of biomolecular interactions without the need for labeling agents. Consequently, refractive index sensing emerges as a potent tool in the realm of label-free biosensing, facilitating a more profound understanding of biological processes and finding applications in diverse fields, including medical diagnostics, pharmaceutical development, and environmental monitoring [4,31].

In this study, we have showcased a comprehensive investigation into a Si₃N₄ platform-based RR device designed to detect refractive index variations in the ambient medium. Our approach integrates both numerical simulations and experimental validations to thoroughly assess the device’s sensing capabilities. Numerical analysis is carried out using the two-dimensional finite element method (2D-FEM) via COMOL Multiphysics software 6.1, offering a reliable approximation of the device’s sensing performance. The significance of this work lies in the remarkable accuracy of the 2D-FEM simulations, which can significantly reduce the computational time. These simulations are so precise that they effectively confirm the experimental validation of the RR structure. This achievement underscores the potential of 2D-FEM simulations in advancing the design and analysis of complex structures, facilitating faster and more cost-effective research and development processes. The fabrication of the photonic sensor involves sophisticated techniques such as electron beam lithography (EBL) and reactive ion etching (RIE) processes. By employing these cutting-edge manufacturing methods, we ensure the precision and reliability of the RR device. Our experimental results align closely with the numerical findings, validating the efficacy of our approach. This research not only establishes the functionality of the device but also highlights the seamless synergy between numerical predictions and real-world performance, underscoring the practical applicability of the Si₃N₄-based RR device for precise refractive index sensing, which can be employed as a label-free biosensor.

2. Results

The effective refractive index (n_eff) of the Si₃N₄ ridge waveguide is meticulously computed across the wavelength spectrum of 1540 nm to 1550 nm, employing a fine step size of 0.1 nm. To ascertain the influence of the waveguide core dimensions on the mode propagating within, variations in the width (W) and height (H) are systematically investigated. Analysis of Figure 1a reveals a discernible trend: the n_eff of the mode exhibits a consistent increase with the enlargement of the waveguide core dimensions. The n_eff of a waveguide is intricately linked to the dimensions of its core. As the dimensions of the core change, the n_eff of the guided modes within the waveguide also undergoes variation. Fundamentally, the n_eff is a measure of the propagation of the speed of light within the waveguide structure relative to the speed of light in a vacuum. When the core dimensions increase, the n_eff tends to rise due to the increased confinement of light within the core material, leading to a longer interaction length between the light and the core. This enhanced interaction promotes a higher n_eff, as more light energy is confined within the core region, resulting in a slower phase velocity compared to the surrounding cladding material. Conversely, decreasing core dimensions can lead to a reduction in the n_eff as the confinement of light weakens, allowing for more energy to spread into the lower refractive index cladding material in the form of an evanescent field.

In consideration of potential fabrication inaccuracies arising from over-etching or under-etching, specific values for the W are judiciously chosen as 1000 nm and 1200 nm, while the H is systematically altered within the range of 340 nm to 370 nm. Importantly, it is emphasized that this range of dimensions is tailored to exclusively support the TE₀ fundamental mode within the specified wavelength range of 1540 nm to 1550 nm. As an illustration, consider the n_eff of the mode within a waveguide characterized by dimensions of W = 1000 nm and H = 340 nm, yielding a value of 1.5267. In contrast, a waveguide with larger dimensions, specifically W = 1200 nm and H = 340 nm, can sustain a fundamental mode with a higher n_eff = 1.5637. This results in superior mode confinement within the core, precluding the potential to design highly sensitive devices because of the decrease in the evanescent field around the waveguide, which could otherwise interact with the ambient medium.

Consequently, formulating photonic sensors with the capacity to provide a more substantial evanescent field is suggested. The evanescent field plays a crucial role in various photonic sensing applications due to its unique property of decaying exponentially outside the core of an optical waveguide. This phenomenon allows the evanescent field to interact sensitively with the surrounding medium, making it a valuable tool for sensing minute changes in the refractive index, concentration, or surface properties of the adjacent materials. The evanescent field can penetrate the analyte, enabling label-free and non-invasive detection methods. The Evanescent Field Ratio (EFR) assumes pivotal significance in the advancement of photonic sensors, particularly those reliant on leveraging the evanescent field for interaction with the surrounding environment. In this context, the evanescent field serves as a crucial element in facilitating the detection and monitoring of variations within the ambient medium [32]. The EFR emerges as a critical parameter because it sums up the intricate connection between the intensity of the evanescent field and its function in detecting dynamic changes.

By precisely evaluating this ratio, photonic sensor designs are crafted to enhance sensitivity and responsiveness, thereby contributing to the refinement and efficacy of sensor technologies in diverse applications, such as environmental monitoring, biomedical diagnostics, and industrial process control. The EFR is defined as the calculated ratio of the integration of intensity within the specified region, typically the upper cladding, to the comprehensive integration of intensity across the entire waveguide structure as follows [33]:

$$\mathrm{E}\mathrm{F}\mathrm{R}=\frac{{\iiint }_{\mathrm{u}\mathrm{p}\mathrm{p}\mathrm{e}\mathrm{r} \mathrm{c}\mathrm{l}\mathrm{a}\mathrm{d}\mathrm{d}\mathrm{i}\mathrm{n}\mathrm{g}}{\left|\mathrm{E}(\mathrm{x},\mathrm{y},\mathrm{z})\right|}^2\mathrm{d}\mathrm{x}\mathrm{d}\mathrm{y}\mathrm{d}\mathrm{z}}{{\iiint }_{\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}}{\left|\mathrm{E}(\mathrm{x},\mathrm{y},\mathrm{z})\right|}^2\mathrm{d}\mathrm{x}\mathrm{d}\mathrm{y}\mathrm{d}\mathrm{z}};$$

(1)

In Figure 1b, the trend becomes evident as the core dimension increases—the n_eff of the mode rises, leading to a concurrent decrease in the EFR. Notably, for TE-polarized light, the width (W) of the core exerts a more pronounced impact on the EFR compared to the height (H) of the core. Therefore, designing a sensing device with a higher EFR value is suggested for enhanced sensitivity.

Furthermore, the EFR has a direct relationship with the sensitivity of the waveguide. To demonstrate this, the sensitivity of Si₃N₄ straight waveguides with different core dimensions is calculated by utilizing the following formula.

$${\mathrm{S}}_{\mathrm{w}\mathrm{a}\mathrm{v}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{i}\mathrm{d}\mathrm{e}}=\frac{∆{\mathrm{n}}_{\mathrm{e}\mathrm{f}\mathrm{f}}}{∆\mathrm{n}};$$

(2)

Here, ∆n_eff represents the variation in the effective index of the mode resulting from a shift in the ambient refractive index (∆n = n₁ − n₂). In this context, we assign values of 1.30 and 1.31 to n₁ and n₂, respectively. The waveguide sensitivity exhibits a decrease from 0.8 n_eff/RIU to 0.5 n_eff/RIU, corresponding to an increase in the core width (W) and height (H). Specifically, this trend is observed as the core dimensions vary from W = 1000 nm (and H = 340 nm) to W = 1600 nm (and H = 370 nm), as illustrated in Figure 1c. To visually elucidate the characteristics of the fundamental mode within the waveguide core with dimensions of W = 1000 nm and H = 340 nm at 1545 nm, refer to the normalized E-field distribution portrayed in Figure 1d. This representation provides valuable insights into the spatial distribution and intensity profile of the electric field associated with the fundamental mode under consideration.

A schematic of the RR structure is presented in Figure 2a, and the transmission spectra of RRs with radii of 50 µm, 75 µm, and 100 µm are meticulously computed across the wavelength spectrum of 1540 nm to 1550 nm with a step size of 0.1 nm, all within the ambient medium of air (n = 1.0). The gap (g) between the bus waveguide and the ring is fixed at 200 nm. Given that the numerical simulations are conducted using the 2D-FEM, the n_eff values for the particular geometry are extracted from Figure 1a. In this context, n_eff values are chosen to pertain to a waveguide geometry with dimensions of W = 1000 nm and H = 340 nm. This selection is pivotal, as it aligns with the subsequent fabrication of the RR structure, ensuring consistency in the geometric dimensions. By employing these n_eff values, the simulations establish a foundation for accurate predictions and assessments in the context of the fabricated RR structure. However, it should be noted that the side wall roughness, sidewall angle, or any imperfections due to fabrication inaccuracy are not considered in the numerical calculations.

To enhance visual clarity and prevent overlap of the resonance dips, the spectra are thoughtfully offset along the Y-axis, as illustrated in Figure 2b. The free spectral range (FSR) is numerically determined for each RR, resulting in values of 4.2 nm, 2.8 nm, and 2.1 nm for radii of 50 µm, 75 µm, and 100 µm, respectively. Furthermore, Figure 2c illustrates the FSR of the RR structure as a function of R within the range of 30 μm to 100 μm. The plot highlights an inverse relationship between R and FSR. This insightful analysis serves as a valuable guide for selecting an optimal R value for the RR structure, aligning with specific FSR requirements. For an in-depth exploration of the resonant behavior, the normalized H-field distribution within the RR with a radius of 100 µm is presented in Figure 2d and Figure 2e for the on-resonance and off-resonance states, respectively. These depictions offer valuable insights into the spatial distribution and intensity of the magnetic field, providing a comprehensive understanding of the RR’s resonant characteristics.

3. Discussion

The transmission spectra of RRs (with upper cladding as air) with R = 50 µm, 75 µm, and 100 µm were meticulously captured within the wavelength range spanning from 1540 nm to 1550 nm, as shown in Figure 3a. The full width at half maximum (FWHM) of the resonance dip at 1543.84 nm is around 0.09 nm, as shown in Figure 3b. Furthermore, the measurement process includes the capture of an image of the RR structure at the resonance wavelength of precisely 1543.84 nm, facilitated by an infrared (IR) camera. At this specific wavelength, a distinct and luminous gleam of light emanates from the ring structure, vividly illustrating its on-resonance state. This captivating phenomenon is visually represented in Figure 3c, where the bright shine within the ring unequivocally signifies its resonance status.

The Q factor of an optical RR is a crucial parameter that describes the efficiency and performance of the resonator in the realm of photonics [34,35]. The Q factor represents the sharpness of the resonant peaks and is defined as follows [36]:

$$\mathrm{Q}\mathrm{f}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{o}\mathrm{r}=\frac{{\mathsf{\lambda }}_{\mathrm{r}\mathrm{e}\mathrm{s}}}{\mathrm{F}\mathrm{W}\mathrm{H}\mathrm{M}};$$

(3)

where λ_res and FWHM are the resonance wavelength and the full width at half maximum of the resonance dip, respectively. A higher Q factor in RRs signifies a more distinct and narrow resonance, making it desirable for applications like optical filters and sensors. The Q factor is affected by material losses, waveguide losses, and radiation losses. Minimalizing these losses is essential for enhancing the Q factor and improving the overall performance of RRs [17]. A high Q factor is essential for optimizing sensitivity and selectivity, enabling diverse applications in integrated photonics. The resonance dip depicted in Figure 3a for R = 100 µm occurs at precisely 1543.84 nm, exhibiting a remarkably narrow FWHM of 0.09 nm. This narrow linewidth translates into an impressive Q factor of 1.7154 × 10⁴, emphasizing the precision and sharpness of the resonant feature.

The refractive index is a fundamental parameter that characterizes how light propagates through a medium, and changes in this index can be indicative of alterations in the composition, concentration, or physical state of the material [27]. In fields such as chemistry, biology, and environmental monitoring, refractive index sensing plays a crucial role in detecting and analyzing substances with exceptional sensitivity [37,38,39]. The ability to exploit minute changes in the refractive index enables advancements in diverse areas, ranging from medical diagnostics to quality control in manufacturing, underlining the versatility and importance of refractive index sensing in advancing scientific understanding and technological innovation [40]. To determine the sensitivity of the fabricated structures, certified refractive index liquids from Cargille Laboratories (series AAA) were used. These liquids are well-described high-quality standard items for sensing and quality control. The droplets of liquids were carefully deposited onto the sample each time and cleaned after the measurements with isopropyl alcohol, making sure that no residue was left. The transmission spectrum of an RR structure with R = 100 µm is plotted in the presence of n = 1.295, 1.305, and 1.315 (±0.001). As the refractive index of the fluid increases, it modifies the n_eff of the propagating mode, resulting in a redshift in the resonance wavelength, as shown in Figure 4a. To optimize the visual clarity and mitigate potential overlap of the resonance dips, the spectra are intentionally offset along the Y-axis, strategically enhancing the distinctiveness of each resonant feature.

The rate of the change in the resonance wavelength versus the refractive index unit (RIU) is plotted in Figure 4b. The slope of the line is estimated by applying a linear regression method. The importance of the linear regression method in establishing the slope of a line cannot be overstated, as it provides a methodical and quantitative framework for comprehending the connection between variables. Through the application of a linear equation to a dataset, this method measures the influence of variations in an independent variable on the dependent variable. The line’s slope, denoted by the coefficient ‘m’ in the equation y = mx + b, assumes a pivotal role as a parameter that signifies the rate of the change in the dependent variable for every unit alteration in the independent variable. The bulk sensitivity of the device is determined as follows [36,41]:

$${\mathrm{S}}_{\mathrm{b}\mathrm{u}\mathrm{l}\mathrm{k}}=\frac{∆\mathsf{\lambda }}{∆\mathrm{n}};$$

(4)

where ∆λ and ∆n are the change in the resonance wavelength and the change in the ambient refractive index, respectively. The device’s sensitivity, obtained through both experimental and numerical computations, stands at 112.5 nm/RIU and 110 nm/RIU, respectively. This minimal difference yields a relative percent error ($\frac{Experiment-numerical}{Experiment}$ × 100) of merely 2.2%, underscoring the precision and reliability of the results.

Table 1. Comparison of numerical and experimental results obtained for RR device.

	FSR (nm)			Sensitivity (nm/RIU)	Q Factor	LOD (RIU)
Radius (µm)	50 ± 0.02	75 ± 0.02	100 ± 0.02
Numerical calculations	~4.2	~2.8	~2.1	~110	-	-
Experimental data	~3.64	~2.42	~1.82	~112.5	1.7154 × 104	7.99 × 10−4
Relative percent error (%)	~15.4	~15.7	~15.3	~2.2	-	-

briefly presents the numerical and experimental findings detailed in this paper, accompanied by the corresponding relative percent errors for comprehensive clarity.

The linearity of an RR sensor refers to its ability to produce a linear response to changes in the parameter it measures, such as the refractive index [42]. In practical terms, this means that as the refractive index of the surrounding medium changes, the resonance wavelength changes in a predictable and proportional manner. The linear correlation coefficient quantifies this relationship mathematically, providing a measure of how closely the sensor’s output follows a straight-line trend with changes in the measured parameter [43]. Our sensing device has a correlation coefficient close to +1, which indicates a strong positive linear relationship, implying that the sensor is highly sensitive and exhibits minimal deviations from linearity over its operating range. Ensuring high linearity and a strong correlation coefficient enhances the sensor’s accuracy and effectiveness in various practical scenarios.

The limit of detection (LOD) within an optical RR-based sensor denotes the smallest discernible change in the measured parameter, such as the refractive index or concentration, that the sensor can reliably identify and can be determined as follows [36]:

$$\mathrm{L}\mathrm{O}\mathrm{D}=\frac{{\mathsf{\lambda }}_{\mathrm{r}\mathrm{e}\mathrm{s}}}{\mathrm{Q}-\mathrm{f}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{o}\mathrm{r}.\mathrm{S}}$$

(5)

where λ_res, Q factor, and S are the resonance wavelength, quality factor, and sensitivity of the device, respectively. Attaining a low LOD holds paramount significance in enhancing the sensor’s sensitivity and its capability to identify subtle variations in the surrounding medium. Various factors influence the LOD of an RR sensor, including intrinsic noise, the signal-to-noise ratio, and the Q factor of the resonator. A higher Q factor, signaling a more distinct resonance and a narrower linewidth, typically results in a lower LOD. The LOD of our sensing device is 7.99 × 10⁻⁴ RIU.

Table 2. Performance comparison of the RR structure fabricated in this work with previously proposed designs.

Platform	Numerical/Experimental	Waveguide Configuration	Sensitivity (nm/RIU)	Q Factor	LOD	Ref.
Silicon nitride	Numerical and experimental	Ridge waveguide	164.8	2575	3.65 × 10−3	[44]
Ge-As-Se chalcogenide	Experimental	Ridge waveguide	123	7.74 × 104	3.24 × 10−4	[45]
Silicon	Numerical	Suspended slot hybrid plasmonic waveguide	458.1	-	3.7 × 10−5	[46]
Polymer	Numerical	Tapered ridge waveguide	84.6 to 101.74	4.6 × 103	-	[47]
Silicon	Numerical	Hybrid double-slot subwavelength grating	1005	22,429	6.86 × 10−5	[48]
Silicon nitride	Numerical and experimental	Ridge waveguide	112.5	1.7154 ×104	7.99 × 10−4	This work

provides a performance comparison of the RR structure fabricated in this work with those reported in previously published papers. It is important to note that this comparison encompasses RR designs utilizing various waveguide configurations, such as ridge, slot, and hybrid plasmonic, across different material platforms, each of which can enhance the sensitivity of the device. The sensitivity and LOD of those from the previous works, as reported in Table 2, are presented in a graphical illustration in Figure 5. This visual representation provides a clear comparison of the performance metrics across different studies, highlighting the need for experimental verification of the studies presented in [44,45,46,47,48] with more advanced waveguide configurations, such as suspended slot hybrid plasmonic waveguides, tapered ridge waveguides, and hybrid double-slot subwavelength grating.

4. Numerical Simulations

The finite element method (FEM) stands as a robust numerical approach widely employed to simulate intricate physical phenomena. COMSOL Multiphysics software harnesses the capabilities of the FEM to tackle a diverse array of multiphysics challenges [49]. Within the COMSOL framework, the FEM partitions a complex simulation domain into smaller, more manageable elements, facilitating the precise representation of complex geometries and material properties. These elements are interconnected at nodes, forming a mesh that discretizes the entire domain.

5. The Fabrication Process

The fabrication process of Si₃N₄ PICs initiates with a precise wafer surface precleaning procedure. To achieve this, we employ established CMOS technology processes—SC-1 and SC-2 [50]. These processes play a pivotal role in eliminating any organic and metallic contamination adhering to the wafer surface, ensuring a pristine starting point for subsequent fabrication stages. Following the precleaning, we embark on two critical processes for material layer formation—wet thermal oxidation and low-pressure chemical vapor deposition (LPCVD). These processes are meticulously designed to yield a 2.3 µm thick SiO₂ layer as the initial foundation, succeeded by a 400 nm Si_xN_y layer. This layering is fundamental for the subsequent stages of the fabrication process.

The subsequent step involves transferring the layout onto a resist, utilizing a 175 nm thick e-beam resist AR-P 6200 from Allresist, employing electron beam lithography (EBL). The exposure process is executed with precision, incorporating proximity effect correction and strategies to minimize stitching errors. The pattern obtained in the resist is then faithfully transferred onto the Si_xN_y layer using dry etching—specifically, reactive ion etching (RIE) with CHF₃/O₂ gas. Any remaining residue on the resist is diligently removed using oxygen plasma. The final fabrication step entails cleaving the wafers into individual chips, culminating in a detailed scanning electron microscope (SEM) inspection. Representative photos illustrating the successful fabrication process are presented in Figure 6. For a more in-depth understanding of the technological processes employed, readers are encouraged to refer to our comprehensive previous work [51]. This work provides a detailed exploration of the intricacies involved in each fabrication step, offering valuable insights into the methodology employed for silicon nitride PIC production.

The inspection unveiled a commendable level of fidelity in reproducing the Si_xN_y shapes concerning the intended layout. The apparent roughness observed on the upper surface of the waveguides is a consequence of the extended etching process. Importantly, this textural irregularity does not impede the optical parameters of the RRs, except for marginally increased optical power losses. The overall integrity of the fabricated components remains intact, underscoring the robustness of the production process despite the surface roughness encountered during etching.

6. The Measurement Procedure

The measurements were conducted with a precise step size of 0.02 nm. To achieve this, tunable laser light was skillfully coupled to the bus waveguide through a tapered lens fiber. The transmitted light was then efficiently collected via an output tapered lens fiber and subsequently directed to the power sensor housed within an Agilent 8163B Lightwave multimeter. The output power, a crucial metric in this analysis, was accurately gauged and visually represented on a computer screen. Facilitating this process was a sophisticated Python script leveraging the PyVISA library. The script seamlessly orchestrated the tuning of the laser source’s wavelength and facilitated the real-time readout of the output power from the power meter. This synergy between the computer and the laser/power meter was made possible through the utilization of general-purpose interface bus (GPIB) equipment. The procedural sequence of the program was as follows: initiating the wavelength setting, incorporating a brief 1 s delay, measuring the output optical power, and culminating in the creation of a comprehensive CSV file encapsulating the acquired data. This well-orchestrated sequence of actions not only ensured precision in the data collection but also streamlined the subsequent analysis and interpretation of the RR transmission spectra. A schematic illustration of the measuring setup is shown in Figure 7.

7. Future Directions for Ring Resonator Structure-Based Sensors

Our team is currently immersed in the numerical analysis and development of Si₃N₄-based ring resonator structures tailored to both sensing and filtering applications. Our aim is to broaden our investigation to encompass highly sensitive devices founded upon waveguide architectures, such as slot waveguides [11], subwavelength grating waveguides [52], and hybrid plasmonic waveguides [53]. By leveraging these architectures, we aspire to achieve the realization of exceptionally sensitive ring resonator-based devices [44,54].

Furthermore, we are poised to undertake the functionalization of photonic sensors for biosensing purposes [55]. This involves intricately modifying the surface of photonic sensors with specific biological molecules or materials to enable the detection and quantification of biological analytes. Such functionalized photonic sensors find widespread utility across diverse biosensing applications, spanning medical diagnostics, environmental monitoring, food safety testing, and drug discovery. Their hallmark attributes include high sensitivity, specificity, and versatility, rendering them indispensable tools in biological and biomedical research endeavors.