Sensitivity Enhancement of Silicon-on-Insulator Multipath Ring Resonator using Gold Nanodisk for Sensor Application

Currently, environmental degradation caused by heavy metals has become a serious concern of many countries. To monitor the concentration of heavy metals in the environment, an in-situ sensor that can measure in real time and has high quality, sensitivity, and flexibility is essential. We proposed a modified multipath ring resonator (MPRR) based on silicon-on-insulator technology with additional gold nanodisk (GND) on top of the ring to increase its sensitivity. To prove the effect of GND on the sensitivity of the modified MPRR, finite-difference time-domain simulations were conducted. Results showed that the average sensitivity of the modified MPRR was 675 nm/RIU, where RIU corresponds to the refractive index unit, higher than that of the unmodified MPRR (171 nm/RIU). Moreover, compared with the single ring structure, the proposed design had better sensitivity. We believe that our proposed approach for the modification of MPRR is suitable for application to optical sensor development.


Introduction
Aside from global warming, environmental problems have currently become a serious concern of society [1]. There are many pollutants in the environment, from nonbiodegradable plastics to dangerous substances, such as heavy metals [1,2]. Thus, daily monitoring of pollutant levels in the environment is critical to ensure that the concentration of these dangerous substances do not go beyond the upper limit. Traditionally, scientists need to obtain samples from the environment (e.g., rivers) and bring these samples to a laboratory. However, this method is considered outdated because sample collection requires time and cannot be done continuously 24 h a day [1]. These conditions led to the need for portable environmental sensors, that is, a compact yet powerful tool to monitor environmental parameters in real time, with good accuracy, high sensitivity, and full compatibility with standard microfabrication facilities for production [3,5].
Most of the commercial environmental sensors sold today are based on electronic signals (e.g., resistive and capacitive sensors), with the drawback of not having exceptional sensitivity [3]. Optical sensors have some advantages over their predecessors, such as high sensitivity, chemically inert, small, and lightweight, suitable for remote sensing, immunity to electromagnetic interference, wide dynamic range, and capable of monitoring a wide range of chemical and physical parameters [3].
Multipath Ring Resonator. The MPRR has a pair of silicon waveguides and two ring resonators, as shown in The inner ring is circular, whereas the outer ring is elliptical. There are four ports, namely, input port (port 1), output resonance (port 3), output antiresonance (port 2), and add port (port 4) [19]. Then, there is the cladding material that encloses the entire area around the ring resonators and silicon waveguides. The cladding material used depends on the analytes of interest [20,21]. How the sensor works is shown in Figure 1.
First, the light enters port 1. An optical phenomenon occurs within the waveguide called an evanescent field. Then, the evanescent field penetrates the cladding material, where the analyte is located. This phenomenon leads to the shifting of the light wave measured in port 3. Figure 2 shows an example of light wave transmittance measured in port 3 and the corresponding refractive index of the cladding material. The red line shows the wave transmitted when the refractive index of the cladding material is 1.00 (in this case, the material used is air). When the refractive index changes, for example, from 1.00 to 1.01, the light wave shifts, as shown by the green line. If the change of the refractive index is high, then the wave will shift far. The sensitivity of the MRR is calculated using the following equation: (1) where S refers to the sensitivity of the MRR, ∆λ represents the shifting of the wavelength, and ∆nclad corresponds to the change of the refractive index in the cladding material [21]. The unit of sensitivity is nm/RIU, where RIU corresponds to the refractive index unit.
Recent studies underlined the optimum configuration of MPRR with decent sensitivity and excellent free spectral range (FSR) and Q factor. However, sensitivity is compromised if the shifting of the light wave exceeds the FSR. This restricts the capability of the MPRR. To solve this problem, the MPRR can be modified by waveguide dispersion [22,23]. Theoretically, this technique will broaden the FSR and consider the importance of adding a metal nanostructure to the structure of the MPRR for sensitivity enhancement. Waveguide dispersion can be accomplished by placing periodic arrays of gold nanodisks (GNDs) above the ring of the MPRR to produce a second-order Bragg grating, which can be expressed as follows: (2) where m = 2 (second order), λB refers to the Bragg wavelength, neff represents the refractive index of a particular mode at λB, and Λ corresponds to the period of the GNDs [22]. Because of these GNDs, interference emanates between the light that enters the ring resonator  [22]. However, this modification scheme that uses GNDs for the MRR has only been explored for the SRR structure. Thus, we investigated the use of this modification scheme for other structures and as the main strategy for sensitivity enhancement.

Design of the Modified Multipath Ring Resonator
In this study, we proposed a modified MPRR based on silicon-on-insulator technology targeted for sensor application. To prove the concept of waveguide dispersion by GNDs in this modified MPRR, 3D finite-difference time-domain (FDTD) simulations were conducted. Table 1 shows several parameters that were adjusted to determine the optimum sensitivity of the modified MPRR.
The optimum thickness of each GND has been optimized from that of a previous report, that is, 30 nm [22]. In this work, we investigated sensitivity enhancement by modifying the diameter of the GND structures and the number of GNDs on the surface of the MPRR. We started from the design of the unmodified MPRR, as shown in the schematic illustrated in Figure 1, that has the following parameters: ring resonator radius (R) = 3 mm, gap separation distance between the silicon waveguide and the ring resonator (G) = 200 nm, gap separation distance between the outer ring and the inner ring (g) = 130 nm, and waveguide width (w) = 450 nm. We set these parameters as the baseline for the modified MPRR. Figure 3(a) shows the schematic of the proposed design of the modified MPRR with GNDs arranged above the rings. The zoomed-in section of the top view of GNDs arranged above the ring, the cross section of the modified MPRR, and the dimensions of a GND are shown in Figures 3(b), 3(c), and 3(d), respectively.
To determine the optimum sensitivity of the modified MPRR, GNDs were configured through several trials. First, the radius of each GND was adjusted from 20 nm to 100 nm. Second, the distance between each GND was adjusted between 0.2 rad and 0.8 rad. We also experimented with GNDs placed above the inner ring only, above the outer ring only, and above both inner and outer rings. Referring to Figure 1, in these simulations, light entered the MPRR through port 1, and the output

b) Zoomed-in Section of the Modified MPRR, (c) Cross-sectional Diagram of the Modified MPRR, and (d) Dimensions of Each GND
was measured in port 3. The transmittance measured in port 3 is similar to that shown in Figure 3, and from this information, the sensitivity was calculated. After the optimum parameter was determined, we compared the sensitivity between modified and unmodified MPRR and between an unmodified SRR structure without GNDs and a modified SRR structure.

Results and Discussion
For the first analysis, we adjusted the radius of the GNDs. Figure 4 illustrates the correlation between the radius of the GNDs and the sensitivity of the MPRR, with a constant value of nclad (i.e., 1.33 RIU). We determined that the maximum sensitivity is reached when the radius of each GND is 87 nm, which is the same radius used in previous work on SRR [22]. The change of sensitivity due to the distance between each GND is shown in Figure 5. Figure 5 illustrates that the maximum sensitivity is reached when the distance between each GND is approximately 0.52 rad.  Figure 6 illustrates how the modification of the GND structure affects the sensitivity of the MPRR by changing the wave shifts. In Figure 6(a), ∆λ and the measured sensitivity are 0.854 nm and 85.43 nm/RIU, respectively. In Figure 6(b), ∆λ and the measured sensitivity are 1.023 nm and 102.39 nm/RIU, respectively. In Figure  6(c), ∆λ and the measured sensitivity are 1.025 nm and 102.59 nm/RIU, respectively. Figures 6(a), 6(b), and 6(c) show the effect of the change of the distance between each GND and its respective ∆λ. The more the GNDs, the lesser the distance between each GND, thus creating a wider ∆λ. Referring to Equation (1), if ∆λ increases, then sensitivity also increases. The results confirmed that our modification scheme involving the change of the diameter of GNDs (Figure 4) and the distance between each GND ( Figure 5) can be effectively used to determine the optimum configuration for the most sensitive designs.
The highest sensitivity is observed when each GND has a radius of 87 nm and there are 72 GNDs, with 36 GNDs on the inner ring and 36 GNDs on the outer ring. This configuration with the highest sensitivity is shown in the schematic illustrated in Figure 3(a), and the sensitivity comparison is presented in Figure 7. Then, we compared the changes of sensitivity when GNDs are placed above the inner ring only, above the outer ring only, and above both inner and outer rings. The results are shown in Table 2.
Then, the modified MPRR is compared with the unmodified MPRR. As presented in Figure 7, the modified MPRR exhibits a higher sensitivity than the unmodified MPRR. We also determined that the sensitivity of the MPRR increases when nclad increases. For this simulation, we increased the value of nclad from 1.00 to 3.30. The maximum sensitivity of the modified and unmodified MPRR is reached when the value of nclad is 3.10. Inversely, the sensitivity of the MPRR decreases afterward with high cavity loss because Si has a refractive index of 3.47 [24]. The maximum sensitivity of the modified MPRR is 3,490.43 nm/RIU. By contrast, the maximum sensitivity of the unmodified MPRR is 1,042 nm/RIU. Additionally, when the refractive index is set to the minimum value of nclad = 1.00 (the cladding material is air), the modified MPRR has slightly better sensitivity than the unmodified MPRR, with the sensitivity of 80.54 and 75.4 nm/RIU, respectively.
For the second analysis, the sensitivity of the modified MPRR is compared with a modified SRR presented in previous research [22]. The sensitivity comparison is shown in Table 3. Model A is the unmodified MPRR, Model B is the modified MPRR, Model C is the unmodified SRR (SRR without GNDs) [22], and Model D is the modified SRR (SRR with GNDs) [21]. These four structures are simulated under similar conditions at nclad = 1.33 RIU (refers to water).   Table 3, we can conclude that the modified MPRR has the highest sensitivity compared with the unmodified MPRR, unmodified SRR, and modified SRR. However, the comparison is limited because this modified MPRR has only been explored for the single ring structure [21,22].

Conclusion
In this study, a modified MPRR with GNDs was proposed and numerically verified. FDTD simulations were conducted to analyze the design and determine the optimum parameter for the modified MPRR. To determine the optimum sensitivity of the modified MPRR, 72 GNDs were required, separated by a distance of 0.52 rad. Every GND has a radius of 87 nm and a thickness of 30 nm. After modification with GNDs, the proposed MPRR reached the maximum sensitivity of 3,490 nm/RIU, and its average sensitivity was 675.43 nm/RIU. Compared with the unmodified MPRR, modified SRR, and unmodified SRR, the modified MPRR has the highest sensitivity when tested under the same conditions at nclad = 1.33 RIU. On the basis of these results, we believe that the proposed MPRR design is highly applicable as a sensor with high sensitivity.
For further research, we propose that real experiments be conducted to prove these simulations. Given the limited resources, the modified MPRR design presented here has only been verified with modeling software. The same goes for the modified SRR that has only been verified by FDTD simulations.