Open Access
Add to BibSonomy
Abstract
Integrated microring resonators are well suited for wavelength-filtering applications in optical signal processing, and cascaded microring resonators allow flexible filter design in coupled-resonator optical waveguide (CROW) configurations. However, the implementation of high-order cascaded microring resonators with high extinction ratios (ERs) remains challenging owing to stringent fabrication requirements and the need for precise resonator tunability. We present a fully integrated on-chip second-order CROW filter using silicon photonic microelectromechanical systems (MEMS) to adjust tunable directional couplers and a phase shifter using nanoscale mechanical out-of-plane waveguide displacement. The filter can be fully reconfigured with regard to both the ER and center wavelength. We experimentally demonstrated an ER exceeding 25 dB and continuous wavelength tuning across the full free spectral range of 0.123 nm for single microring resonator, and showed reconfigurability in second-order CROW by tuning the ER and resonant wavelength. The tuning energy for an individual silicon photonic MEMS phase shifter or tunable coupler is less than 22 pJ with sub-microwatt static power consumption, which is far better than conventional integrated phase shifters based on other physical modulation mechanisms.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Photonic integrated circuits (PICs) are a state-of-the-art technology platform that can provide energy-efficient solutions for high-speed signal processing. Due to the high degree of integration and significant energy-efficiency gains [1], PICs are the subject of active research in quantum photonics [2], light detection and ranging (LiDAR) sensors [3], and programmable photonic circuits [4,5]. In these research directions, a common challenge is the need to scale PICs to large-scale circuits (with potentially >106 photonic components on a single chip [6]). Additionally, individual devices are susceptible to nanoscale manufacturing variations. To counter these variations, active control of the optical signal by phase shifters can be implemented, typically based on the thermo-optic effect [7–9], phase-change materials (PCM) [10,11], ferroelectric domain poling [12], the piezoelectric effect [13,14], and microelectromechanical systems (MEMS) [15–20]. However, most thermo-optic devices exhibit high power consumption ranging from tens to hundreds of milliwatts per device in the conventional silicon- and silicon nitride-based PIC platforms, and it is challenging to control the temperature and thermal crosstalk for large device counts. Both PCM and ferroelectric domain poling have been demonstrated for nonvolatile photonic phase control; however, the number of actuation cycles have been limited, and insertion losses are still significant for further scaling. Piezoelectric tuning has been demonstrated for energy-efficient phase control. However, owing to the weak optical effect, the footprint of individual phase shifters poses challenges in scaling this technology.
In contrast, integrated silicon photonic MEMS provide a unique combination of low power consumption, small optical losses, and compact footprints, making them suitable for scalable, energy-efficient data communication applications or optical signal processing. Consequently, advanced MEMS-based PICs, such as tunable add–drop filters based on coupled-resonator optical waveguides (CROWs), are expected to play a key role in integrated photonic signal processing. CROW-based photonic signal processing architectures have been commonly employed in nonlinear optics and communication systems [21]. They allow for shaping of the spectral filter response with high-order multiple coupled microring resonators through fine phase control of each individual resonator. Furthermore, by employing cascaded CROW with varied free spectral ranges (FSRs), we anticipate enhanced performance compared to a single resonator or a CROW where all resonators have the same FSR, such as the Vernier effect to increase the FSR of the cavity. However, nanoscale fabrication tolerances may cause deviations from the target resonant frequencies or coupling coefficients. This susceptibility to fabrication tolerances makes simultaneous and independent control of both high extinction ratio (ER) and phase tuning in a single device challenging, despite the well-established understanding of CROW-based optical filtering techniques [22]. MEMS can provide a unique solution to adjust both the phase and coupling, as recently demonstrated for a single-ring resonator device [18].
In this study, using silicon photonic MEMS, we experimentally demonstrate a fully reconfigurable add–drop filter based on a second-order CROW configuration, which allows for precise control of two key parameters: (1) the coupling coefficients between two ring resonators and between the ring and bus waveguide, and (2) the resonant frequency of one of the ring resonators. Our demonstration suggests the potential to compensate for fabrication tolerances and achieve energy-efficient reconfiguration with ultra-low power consumption.
2. Design and simulation
Figure 1(a) shows a three-dimensional schematic of the proposed MEMS-based CROW device. The four ports (In, Through, Add, Drop) are defined as shown in the diagram. The second-order CROW consists of two ring resonators and two bus waveguides (Fig. 1(b)). The bus waveguides were connected to the grating couplers for coupling with optical fibers. The ring resonators and waveguides are optically coupled via MEMS-tunable directional couplers, and one of the ring resonators features a MEMS-tunable phase-shifter. The round-trip lengths of the two ring resonators are designed to have distinctly different FSRs. The round-trip lengths are 5488 µm (ring 1) and 4488 µm (ring 2), and the corresponding FSRs (named FSR1, FSR2) are 0.103 and 0.126 nm, respectively, near an operation wavelength of 1550 nm. The phase shifter is located at the shorter ring (ring 2) resonator.
Fig. 1. (a) Three-dimensional schematic of the second-order CROW add–drop filter (not to scale). (b) Optical microscope image of the fabricated device with grating couplers. The ring resonators are marked with red lines (ring 1, ring 2), and the bus and perturbing waveguides are marked with blue and green lines, respectively. (c) Schematics of the MEMS-tunable directional couplers representing three major coupling states (the red line indicates the path of the guided light from the bus waveguide). (d) Simulated optical response of the MEMS-tunable directional coupler versus the vertical offset.
The waveguides used in the device are 220 nm thick and have a 70-nm-thick slab on one side to provide mechanical support (Fig. 1(a) inset). Two types of waveguides are used in the device: the main signal-carrying waveguide (represented as blue and red colors in Fig. 1(b)) and the field-perturbing waveguide (represented as green in Fig. 1(b)).
The main waveguide has a width of 450 nm and is used for light guiding and tunable coupling. It is designed to be operated in the transverse-electric (TE) mode near a wavelength of 1550 nm, and its simulated effective and group indices are 2.3257 and 4.2558, respectively. The perturbing waveguide has a width of 300 nm. A perturbing waveguide is used in the MEMS-tunable phase shifter to change the effective index of the optical mode guided in the main waveguide to create a phase-shifting effect.
We use electrostatic cantilever actuators to control tunable couplers and phase shifters. The cantilevers are 220 nm thick, and they are initially elevated by the film stress of the Cr/Au deposited at their hinges (Fig. 1(a)). The tip of the cantilever is lifted by >1 µm when no voltage is applied. The actuators are moved vertically by the electrostatic force between the actuators and the silicon substrate when voltages are applied to the actuators. By changing the voltage applied to the actuators, we can control the actuator displacements and therefore the states of the tunable couplers and phase shifters.
Figure 1(c) shows the operating principle of the MEMS-tunable directional coupler. The coupler consists of two parallel waveguides with the identical core width, and one of the waveguides is attached to the cantilever actuator. By changing the vertical offset of the cantilever tip and therefore the vertical position of the attached waveguide, the optical coupling between the two waveguides can be continuously changed. Thus, the coupling ratio of the tunable directional coupler can be tuned. The straight coupling section of the directional coupler is 200 µm long. The lateral gap between the waveguides is 250 nm when the cantilever actuator is in the in-plane position (vertical offset = 0 nm). The simulated optical response of the tunable directional coupler is shown in Fig. 1(d). The optical simulation was conducted using the Lumerical Finite Difference Eigenmode (FDE) solver. The coupling ratio of the directional coupler can be tuned in the range of 0 to 1 by changing the vertical offset from 1000 to 300 nm.
Figure 2(a) illustrates the cross sections of the cantilever-based MEMS-tunable phase shifter. The phase shifter consists of a main waveguide (width = 450 nm) and a perturbing waveguide (width = 300 nm). The light is mainly guided by the main waveguide, whereas the perturbing waveguide perturbs the optical mode in the main waveguide to change its effective refractive index [20]. The narrow design prevents mode excitation and leads to reduce the refractive index and the low index prevents the optical power leakage. The perturbing waveguide is 1875µm long and is attached to a cantilever actuator. When the actuator is in the in-plane position (vertical offset = 0 nm), the lateral gap between the main waveguide and the perturbing waveguide is 200 nm. By pulling the cantilever actuator and therefore the vertical position of the perturbing waveguide, the amount of perturbation to the optical mode in the main waveguide can be changed. In this manner, the effective refractive index of the optical mode can be continuously modified, changing the phase of the light passing through the phase shifter. Figure 2(b) shows the simulated optical response of the phase shifter. The effective refractive indices of the optical mode were simulated using the FDE solver, and the amount of phase shift was calculated from the effective refractive index variation using the following equation:
where L, $\mathrm{\Delta }{n_{eff}}$, $\lambda $, and $\mathrm{\Delta }\phi $ represent the length of the perturbing waveguide, effective refractive index change, operating wavelength (i.e., 1550 nm), and amount of phase shift, respectively. The effective refractive index of the optical mode changed from 2.3257 to 2.3265($\mathrm{\Delta }{n_{eff}}$=0.0008) when the vertical offset changed from 1000 to 0 nm, and a full 2π phase shift was achieved. During the fabrication process, the dimension of the waveguide can affect the amount of phase shift. Specifically, as the waveguide dimension decreases, the phase shift increases as shown in Fig. 2(c). Additionally, as the wavelength increases, the phase shift amount can also increase since the effect of $\mathrm{\Delta }{\textrm{n}_{\textrm{eff}}}\textrm{}$ is greater than the change of wavelength near 1550 nm.Fig. 2. (a) Cross-sectional view of the MEMS-tunable phase shifter and mode profile of each state. Simulated response of the MEMS-tunable phase shifter (b) with the desired dimension and (c) depending on the fabrication variation.
The fundamental mechanical resonant frequencies of the cantilever actuators used in the tunable coupler and phase shifter can be calculated as follows [23]:
where h represents the thickness of the cantilever beam, w represents the beam length, $\rho $ represents the mass density of the beam, and E represents the Young’s modulus of silicon. the cantilevers used in the tunable coupler and phase shifter have the same beam length ($w$) and thickness ($h$), they share the same resonant frequencies. The resonant frequency of the first mode was calculated as 156 kHz with $E = 160\; \textrm{GPa}$, $w = 43.45\; {\mathrm{\mu} \mathrm{m}}$, $h = 220\; \textrm{nm}$, and $\rho = 2.329\; \textrm{g}/\textrm{c}{\textrm{m}^3}$. Because the actuator’s response time is inversely proportional to the resonant frequency ($t\; \sim \; 1/{f_0})$, we can expect a microsecond-scale response time.3. Device fabrication
The MEMS devices were fabricated at the National NanoFab Center (NNFC) in South Korea using a process similar to that of standard passive silicon photonics. Figure 3(a) shows a cross-sectional schematic of the fabricated device. The device was fabricated on an 8-inch silicon-on-insulator (SOI) wafer with a 220-nm-thick silicon device layer and 2-µm-thick BOX layer. A 100-nm-thick silicon dioxide layer was deposited on top of the silicon layer as a hard mask. Three lithography steps were performed to define the silicon layer using a KrF scanner. Each step consisted of 70-, 80-, and 70-nm-deep silicon etching to define etch hole in MEMS structure, waveguide, and grating coupler. The etch holes are strategically designed as 2 µm squares, arranged in an array with equal spacing of 3 µm, aimed at completely removing oxide under the cantilever during the VHF process. After the silicon layer was defined, an additional lithography step was performed with an I-line stepper to form a liftoff mask for the metal-layer definition. The metal layer was evaporated and consisted of 5-nm-thick chromium (for adhesion and mechanical stress) and 30-nm-thick gold. After metallization, VHF etching was used to undercut the BOX layer to release the MEMS actuators and waveguides. Figures 3(b)–(d) show scanning electron microscopy (SEM) images of the fabricated device. The total area of the device, including the grating couplers, was <2.3 mm2. The areas of the metal pads for electrical probing and wire bonding were 70 × 215 and 70 × 1890 µm2 for the tunable couplers and the phase shifter, respectively. In Fig. 3(d), the waveguides constituting the tunable coupler are colored for clarity.
Fig. 3. (a) Cross-sectional schematic of the fabricated wafer. The silicon device layer and buried oxide (BOX) layer were 220 nm and 2 µm thick, respectively. Cr/Au was deposited as a probing pad and stress layer for the micromechanical hinge. There were 2-µm-wide periodic etch holes in the silicon layer for the vapor hydrofluoric acid (VHF) process. The widths of the main and perturbing waveguides were 450 and 300 nm, respectively. (b) SEM image of the fabricated chip. (c) SEM image of the tunable coupler. (d) Magnified SEM image of the movable waveguide on the tunable coupler and the fixed bus waveguide. The waveguides are marked with blue line.
4. Spectral response measurement
For experimental measurement, A tunable laser source (Toptica CTL 1550) centered at a wavelength of 1550 nm was used as the light source. A fiber array was used to couple the light in/out from the grating couplers on the chip. Photodetectors with a 30-dB dynamic range were connected to an oscilloscope and used to record the optical responses of the device. A fiber Mach–Zehnder interferometer (MZI) with a 100-MHz FSR was used to precisely record the relative wavelength of the light source while tuning the wavelength. Tungsten probe tips were used to probe the metal pads and bias actuators. DC power supplies were used for static measurements, and a function generator was used to generate waveforms for dynamic measurements.
Figure 4(a) shows the measured optical transmission spectra of ring 2 with different voltages applied to the MEMS-tunable coupler between the In and Through ports as shown in Fig. 1(a). As the coupling ratio of the tunable coupler was increased by increasing the applied voltage, the ERs of the resonance dips increased. An ER of 25 dB was achieved when a bias voltage of 13.8 V was applied to the coupler actuator. The measured FSR of ring 2 was FSR2 = 0.123 nm, which was slightly smaller than the simulated value of 0.126 nm. The difference may have resulted from the fabrication variation of the waveguide, which resulted in a shift in the group index of the waveguide. When we increased the voltage applied to the tunable coupler, we observed a coupling-induced resonance frequency shift [24]. This shift can be compensated for by the MEMS-tunable phase shifter attached to ring 2.
Fig. 4. Optical characterization of Ring 2 with tuning of the MEMS-tunable directional coupler and a phase shifter. (a) Measured transmission spectrum of the single ring resonator depending on the voltage applied to the MEMS-tunable directional coupler. (b) Measured transmission spectrum of the single-ring resonator depending on the voltage applied to the MEMS-tunable phase shifter. (c) Measured resonance peak shift from (b) versus the voltage applied to the phase shifter. (d) One resonance dip of the measured spectrum (red line) and the fitted Lorentzian function (black dotted line).
Figure 4(b) shows the measured optical transmission spectra between the In and Through ports with different voltages applied to the MEMS-tunable phase shifter attached to ring 2, utilizing the same tunable coupler used in Fig. 4(a). When 10 V was applied to the phase shifter, the transmission spectrum shifted by 0.135 nm, which was larger than FSR2. Figure 4(c) shows the measured spectrum shift (left y-axis) as a function of the voltage applied to the phase shifter. The spectrum shifted by >0.195 nm at 12 V. The amount of phase shift induced by the phase shifter was determined according to the amount of spectrum shift and indicated in the same graph (Fig. 4(c), right y-axis). As shown in the graph, a phase shifter can induce a phase shift exceeding 3π. By fitting resonance dip of the spectrum with a Lorentzian function, as shown in Fig. 4(d), and assuming a state of critical coupling, we determined the intrinsic quality factor of the ring to be 2.32 × 105 since this assumption entails that the intrinsic quality factor is twice the value of the loaded quality factor, which was measured 1.16 × 105 from the FWHM and center wavelength.
Using our device, we configured a second-order CROW filter (Fig. 5(a)) and measured its spectral response. Figure 5(b) shows the measured optical transmission spectra at the Through and Drop ports of the second-order CROW filter when the input light was coupled into the In-port. The simulated spectra of the CROW filter overlap with the measured spectrum in the graph. The simulation was performed using Lumerical Interconnect software. The FSRs of each ring resonator were measured as 0.100 nm (FSR1) and 0.123 nm (FSR2), as shown in the graph. Figure 5(b), observations indicate that certain resonances remain distinct, while others converge into a unified central wavelength within the CROW configuration. Individual 3 dB bandwidths were recorded at 2.72 GHz and 2.08 GHz for Ring 1 and Ring 2, respectively. In contrast, the second-order CROW exhibited a bandwidth of 3.83 GHz. An examination of the roll-off rates revealed measurements of 3.24 dB/GHz and 3.00 dB/GHz for Ring 1 and Ring 2, respectively, while the second-order CROW presented a significantly higher rate of 5.6 dB/GHz. These results demonstrate the feasibility of creating a bandpass filter with CROW compared with a single ring resonator [25]. By employing a MEMS-based phase shifter, we can manipulate the roll-off rate of this filter, pointing out the advantages of the CROW system when compared to a single microring resonator. Figure 5(c) shows the drop-port spectra of the second-order CROW filter. Initially, we merged the two resonance peaks to obtain a flatter passband by applying 13.08 V to the phase shifter attached to ring 2. By raising the voltage to 13.3 V, the resonance peaks of the two rings were clearly distinguishable. This relative wavelength, contingent on the voltage applied to the phase shifter, was determined by setting the wavelength of resonance 2 as the baseline at zero before it shifted. Figure 5(d) shows the evolution of the resonance wavelengths of the two peaks in Fig. 5(c) during a sweep of the voltage applied to the phase shifter. The data corresponding to 13.08 V and 13.3 V in Fig. 5(c) are situated at the 17th and 20th positions, respectively in Fig. 5(d). The two peaks become closer near 13 V, and we can clearly observe the resonance anti-crossing phenomenon in the graph. Our device architecture can be used to study the dynamics of photonic molecules [26,27].
Fig. 5. Measured optical spectrum of the second-order CROW configured with the MEMS-tunable couplers and the phase shifter. (a) Configuration of the measured CROW. (b) Measured and simulated transmission spectra of the drop and through ports of the CROW when a tunable light source was coupled into the in-port. The black and red lines indicate the experimental data of the through and drop ports, respectively, and the blue and green lines indicate the simulated data of the through and drop ports, respectively. (c) Measured transmission spectra of the drop port while the phase shifter was in operation. (d) Locations of resonance peaks measured during the tuning of the phase shifter. Splitting of the two resonance peaks near 13 V is clearly observed.
5. Mechanical and electrical characterization
Figure 6(a) presents the measured time response of the tunable coupler in response to step bias voltages. The voltage interval was set to achieve a transmission difference of 20 dB. As shown, the rise and fall times were measured as 8.72 and 5.79 µs, respectively. The rise and fall times were defined as the times required to reach 90% and 10% of the maximum optical signal, respectively, in response to the step voltage change. Figures 6(b) and (c) show the measured Bode magnitude and phase plot, respectively, of the optical response of the tunable coupler in response to sinusoidal waveform voltages. To assume that the mechanical displacement and optical response are linearly proportional, the maximum and minimum values of the sinusoidal waveform voltage were set within the range that can operate within the linear regime of the tunable directional coupler as shown in Fig. 6(a). The black lines in the graphs indicate the fitted curves based on the following standard response equation:
where $\zeta $ represents the damping ratio, and ${f_0}$ represents the resonant frequency of the system. We found the values of $\zeta $ and ${f_0}$ to be 1.187 and 96.6 kHz, respectively. We attribute the discrepancy between the calculated mechanical resonance frequency (156 kHz) and the measured resonance frequency (96.6 kHz) to the electrical RC delay time. The graphs indicate that only one dominant mechanical mode was measured under the operation frequency range below 1 MHz. This result can be attributed to the RC delay, which functions similarly to a low-pass filter. Consequently, although other modes might be present, we could reliably drive the system without needing to shape the waveform of the applied voltage or reduce the application speed. Moving forward, if we can overcome the RC constraints through silicon doping, it would become feasible to observe all the mechanical modes.Fig. 6. Mechanical characterization of the MEMS-tunable directional coupler. (a) Measured time response of the tunable coupler. Measured frequency response of the tunable coupler: (b) magnitude plot and (c) phase lag plot.
An electrical parameter analyzer (Keithley 4200A-SCS) was used to measure the electrical power and energy consumption of the device. Figure 7(a) shows the measured current response of the MEMS-tunable directional coupler when a step voltage of 9 V was applied to the actuator. By multiplying the measured instantaneous current by the applied voltage, we obtained an electrical power consumption graph (Fig. 7(b)). We obtained the energy needed to turn the coupler on at 9 V, i.e., 8.83 pJ, by calculating the area under the curve in the transient state. Similarly, we obtained the tuning energies required for the tunable coupler (Fig. 7(c)) and phase shifter (Fig. 7(d)) at several operation voltages, with quadratic fitting functions. The maximum energy consumptions of the tunable coupler and the phase shifter were 21.2 pJ (at 13.8 V, maximum coupling condition) and 11 pJ (at 9.75 V, 2π-phase shift condition). These values are several orders of magnitude lower than the tuning energy for thermo-optic tuning (i.e., nJ–µJ). The energy needed to operate the coupler mostly originates from the electrical energy to charge the capacitance of the electrostatic MEMS actuator. Even though no electrical power consumption is expected in the steady state, there is a small finite leakage current through the insulating layer of the actuator (i.e., 2-µm-thick BOX). The minimum measurable power of our setup was 100 nW, and the leakage was below the minimum value.
Fig. 7. Electrical characterization of the MEMS-tunable directional coupler and the phase shifter. (a) Measured current and voltage applied to the tunable coupler. (b) Measured electrical power consumption of the tunable coupler. Measured tuning energy and fitted quadratic curve of the (c) tunable coupler and (d) phase shifter. Owing to the limitations of our measurement setup, measurements were only performed up to a voltage of 9 V.
To highlight the novelty of this research, we have executed a comprehensive comparative analysis between the performance of our work and other devices, both utilizing the same silicon photonic MEMS technology for ring resonator filter. The details of which are presented in Table 1. Recently, the device delineated in [18] shows the tuning of both directional coupler and phase shifter that differs from others, but it poses scalability challenges due to its structure, particularly when an expansion to larger orders is envisaged. Notably, we are the first demonstrate cascaded microring resonator featuring reconfigurability in both coupling strength and phase tuning that shows the possibility to expand a larger-scale CROW filter.
Table 1. Performance comparison among silicon photonic MEMS-based tunable ring resonators
6. Conclusion
We designed and experimentally demonstrated a MEMS-based second-order CROW optical filter with full tunability and ultralow tuning energy. The device can tune the resonance wavelength and coupling strength between the ring resonators and bus waveguides. Each tuning element in the device consumes <22 pJ of electrical energy for reconfiguration. The static power consumption of the elements was <100 nW, and their reconfiguration time was <10 µs. We believe that our device architecture can be used to implement higher-order CROW filters with flexible reconfiguration capabilities and ultralow tuning energies.
Funding
Agency for Defense Development (UI210008TD).
Acknowledgments
Portions of this research were previously disseminated at CLEO 2021 in paper STh1Q.5 [28].
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. D. Thomson, A. Zilkie, J. E. Bowers, et al., “Roadmap on silicon photonics,” J. Opt. 18(7), 073003 (2016). [CrossRef]
2. J. Wang, F. Sciarrino, A. Laing, et al., “Integrated photonic quantum technologies,” Nat. Photonics 14(5), 273–284 (2020). [CrossRef]
3. B. Behroozpour, P. A. M. Sandborn, M. C. Wu, et al., “Lidar System Architectures and Circuits,” IEEE Commun. Mag. 55(10), 135–142 (2017). [CrossRef]
4. D. Pérez-López, A. López, P. DasMahapatra, et al., “Multipurpose self-configuration of programmable photonic circuits,” Nat. Commun. 11(1), 6359 (2020). [CrossRef]
5. W. Bogaerts, D. Pérez, J. Capmany, et al., “Programmable photonic circuits,” Nature 586(7828), 207–216 (2020). [CrossRef]
6. N. Margalit, C. Xiang, S. M. Bowers, et al., “Perspective on the future of silicon photonics and electronics,” Appl. Phys. Lett. 118(22), 220501 (2021). [CrossRef]
7. Z.-W. Li, Q. Liu, H. Wang, et al., “Photo-Induced Thermo-Optical Refraction Switching by a Graphene-Assisted Silicon Microring Resonator,” J. Lightwave Technol. 39(11), 3471–3477 (2021). [CrossRef]
8. S. Chung, M. Nakai, and H. Hashemi, “Low-power thermo-optic silicon modulator for large-scale photonic integrated systems,” Opt. Express 27(9), 13430 (2019). [CrossRef]
9. W. Zhang and J. Yao, “Photonic integrated field-programmable disk array signal processor,” Nat. Commun. 11(1), 406 (2020). [CrossRef]
10. Z. Fang, R. Chen, J. Zheng, et al., “Ultra-low-energy programmable non-volatile silicon photonics based on phase-change materials with graphene heaters,” Nat. Nanotechnol. 17(8), 842–848 (2022). [CrossRef]
11. R. Chen, Z. Fang, F. Miller, et al., “Opportunities and Challenges for Large-Scale Phase-Change Material Integrated Electro-Photonics,” ACS Photonics 9(10), 3181–3195 (2022). [CrossRef]
12. J. Geler-Kremer, F. Eltes, P. Stark, et al., “A ferroelectric multilevel non-volatile photonic phase shifter,” Nat. Photonics 16(7), 491–497 (2022). [CrossRef]
13. W. Jin, R. G. Polcawich, P. A. Morton, et al., “Piezoelectrically tuned silicon nitride ring resonator,” (n.d.).
14. J. P. Epping, D. Marchenko, A. Leinse, et al., eds. (2017), p. 101060F.
15. S. Han, T. J. Seok, N. Quack, et al., “Large-scale silicon photonic switches with movable directional couplers,” Optica 2(4), 370 (2015). [CrossRef]
16. C. Errando-Herranz, F. Niklaus, G. Stemme, et al., “Low-power microelectromechanically tunable silicon photonic ring resonator add-drop filter,” Opt. Lett. 40(15), 3556 (2015). [CrossRef]
17. H. M. Chu and K. Hane, “A Wide-Tuning Silicon Ring-Resonator Composed of Coupled Freestanding Waveguides,” IEEE Photon. Technol. Lett. 26(14), 1411–1413 (2014). [CrossRef]
18. P. Edinger, G. Jo, C. P. Van Nguyen, et al., “Vacuum-sealed silicon photonic MEMS tunable ring resonator with an independent control over coupling and phase,” Opt. Express 31(4), 6540 (2023). [CrossRef]
19. H. Sattari, A. Y. Takabayashi, P. Edinger, et al., “Silicon photonic microelectromechanical systems add-drop ring resonator in a foundry process,” J. Optical Microsystems 2(04), 6540 (2022). [CrossRef]
20. P. Edinger, A. Y. Takabayashi, C. Errando-Herranz, et al., “Silicon photonic microelectromechanical phase shifters for scalable programmable photonics,” Opt. Lett. 46(22), 5671 (2021). [CrossRef]
21. L. Caspani, C. Xiong, B. J. Eggleton, et al., “Integrated sources of photon quantum states based on nonlinear optics,” Light: Sci. Appl. 6(11), e17100 (2017). [CrossRef]
22. J. Poon, J. Scheuer, S. Mookherjea, et al., “Matrix analysis of microring coupled-resonator optical waveguides,” Opt. Express 12(1), 90 (2004). [CrossRef]
23. R. F. Gibson, Principles of Composite Material Mechanics, 3rd ed., Dekker Mechanical Engineering (Taylor & Francis, 2012).
24. M. Popovic, C. Manolatou, and M. Watts, “Coupling-induced resonance frequency shifts in coupled dielectric multi-cavity filters,” Opt. Express 14(3), 1208 (2006). [CrossRef]
25. Y. Liu, Y. Yu, L. Wang, et al., “Reconfigurable Microwave Photonic Bandpass Filter Based on CROW,” J. Lightwave Technol. 14, 1–8 (2023). [CrossRef]
26. B. Peng, Ş. K. Özdemir, J. Zhu, et al., “Photonic molecules formed by coupled hybrid resonators,” Opt. Lett. 37(16), 3435 (2012). [CrossRef]
27. M. Zhang, C. Wang, Y. Hu, et al., “Electronically programmable photonic molecule,” Nat. Photonics 13(1), 36–40 (2019). [CrossRef]
28. Y. J. Park, D. U. Kim, D. Y. Kim, et al., “Fully Reconfigurable Coupled-Resonator Optical Waveguides (CROWs) with 10 nW Static Power MEMS,” in Conference on Lasers and Electro-Optics (Optica Publishing Group, 2021), paper STh1Q.5.
