Open Access
Add to BibSonomy
Abstract
Dispersion impairments are a well-known limitation in data center communications, limiting both the usable data rates and reaches. Several companies today adopt silicon photonics as a core technology in their transceiver products. This presents an opportunity for silicon photonics-based dispersion management technologies to be integrated with the transceiver transmitter or receiver. In this manuscript, we present a ring-resonator based, tunable dispersion compensation device, providing dispersion ranging as wide from + 12.9 × 103ps/nm to −12.3 × 103ps/nm. Thermo-optic tuning from 20°C to 70°C is demonstrated to allow continuous wavelength tuning across 200 GHz. High-speed experiments using 25 Gb/s non-return-to-zero data propagating through 20 km of single mode fiber show that a significant improvement in the eye diagram is achieved after compensation with the ring-resonator device. We demonstrate a significant improvement in the BER from 10−3 to 10−11 for data rates of 25 and 25.78125 Gb/s.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Data center communication relies critically on error-free transmission of high speed data [1,2]. The propagation of light fields in optical media is subject to chromatic dispersion, which arises from the frequency dependence of the effective refractive index. With higher transmission rates and longer reaches, the impact of dispersion induced impairments become more pronounced [3–5]. Transceiver companies have reported limitations in both the transceiver reach and data rate as a result of dispersion impairments, including those which utilize Non-Return-to-Zero (NRZ) [6,7] and Pulse Amplitude Modulation 4 levels (PAM4) modulation formats [6–11]. Fortuitously, dispersion is a linear phenomenon. Therefore, mitigating dispersion may be performed by concatenating another device with dispersion equal in magnitude but opposite in sign. Consequently, the deployment of transceivers with dispersion compensation in data center communications may overcome both reach and data rate limits imposed by the dispersion inherent to the transmitting fiber.
Dispersion compensation manages dispersion by combining the dispersive effects from two or more photonic components such that the aggregate Group Velocity Dispersion (GVD) of the entire optical link is low [12]. Dispersion compensation techniques can be used at the transmitter end, along the transmission channels or with pre-compensation at the laser source [12,13]. With the widespread deployment of silicon-photonics based transceivers, CMOS-compatible devices that can seamlessly be integrated with the rest of the transceiver chip are therefore compelling technologies. Perfect cancellation of chromatic dispersion introduced by the transmitting fiber requires a component with a dispersion equal in magnitude and opposite in sign. Furthermore, the ability to tune the dispersion magnitude is of great merit. Consequently, devices which are subjected to temperature fluctuations from the environment, are susceptible to finite fabrication tolerances and material inhomogeneities which could result in a certain extent of variation in the exact lasing wavelength of laser source. Hence, tunable dispersive devices also have the added advantage to compensate for different dispersion characteristics to accommodate manufacturing variations.
In this manuscript, we demonstrate on-chip dispersion compensation using a CMOS-compatible silicon nitride (Si3N4) ring resonator using the drop port. Thermo-optic tuning is shown to result in a shift in the resonant wavelength of ∼34pm/°C. We also characterize the dispersion at the drop-port of the ring resonator to vary from as high as $+ 12.9 \times {10^3}ps/nm$ to $- 12.3 \times {10^3}ps/nm$ with a temperature change of ±0.5°C. High-speed experiments are performed using 25 Gb/s NRZ data to show that the dispersion arising from 20 km of optical fiber can be substantially restored after dispersion compensation using the Si3N4 ring resonator.
2. Ring resonator design
Ring resonator devices are intrinsically highly wavelength selective. As a result, the optical path length traversed by a specific wavelength of light varies rapidly close to each resonance, resulting in large dispersion. The drop-port from the ring resonator design can be used to precisely compensate for dispersion that can be tuned by varying the temperature. The ring resonator has a waveguide width (W), height (H), radius (R) and coupling gap (G) of 1.5µm, 800 nm, 50µm and 400 nm respectively (Fig. 1(a)). With the coupling of two waveguides to the ring resonator, some extent of the incident field or input wave is coupled to the drop port, while the remaining incident field is transmitted to the through port. Transmission through both ports after coupling with the ring resonator can be described by [12]:
where $U({z,t} )$ is the normalized amplitude, z is the propagation direction and β2 is the GVD. Dispersion compensation would require the dispersion in the compensation device to match that of a second device to achieve an aggregate dispersion that is close to zero. The combined dispersion of the optical path may then be described by [12]:
3. Optical characterization of the ring resonator dispersion
We utilize a dispersion analyzer which interferometrically extracts the phase information from the ring resonator devices. The phase is then used to extract the dispersion properties. The transmission spectra are measured using a tunable continuous wave laser and synchronized detector. For the thermo-optic tuning of the device, a temperature controller with a Peltier module was used as a heat source to vary the temperatures. The device was placed entirely on the Peltier module before the alignment of a tapered fiber coupler to the device under test. A tapered lensed fiber was used to facilitate coupling between the ring resonator and fiber. In all experiments, a polarization controller with 20 dB of polarization extinction was used to select the transverse-electric polarization prior to coupling into the devices.
Figure 2(a) shows the transmission spectrum of the ring resonator at the through and drop ports at a temperature of 20°C for the quasi-TE mode. We note that the resonance arising from the quasi-TM mode may be seen in the spectrum due to the limited polarization extinction ratio of the polarization controller used in the experiments. Figure 2(b) shows the close-up view of a TE-mode resonance at 1549.13 nm. The group delay at the drop-port was extracted by first measuring the phase information as a function of wavelength. Thereafter, the group delay may be calculated using the derivative of the phase shift $\varphi (\omega )$ with respect to the frequency, using the expression $\tau (\omega )= \frac{{\partial \varphi (\omega )}}{{\partial \omega }}$. It may be observed from Fig. 2(b) that the peak group delay at the resonance located at 1549.13 nm is 3.29 ns. The dispersion may then be extracted using the derivative of the measured group delay, as shown in Fig. 2(c). It is observed that the dispersion at the drop port varies rapidly with the wavelength detuning, ranging from being strongly anomalous on the blue-side to strongly normal on the red-side of the drop port resonance.
Fig. 2. (a) Transmission spectrum of the through (red) and drop (blue) ports as a function of wavelength. (b) Transmission (blue), group delay (red) and (c) dispersion (fuchsia) measured at 20°C for the resonance close to 1550 nm. (d) Dispersion as a function of temperature tuning.
Standard single mode fiber used in data center communications possesses a dispersion of 16ps/nm/km at 1550 nm. Consequently, for compensation of a single mode fiber, the resonator needs to be operated with a red-detuning with respect to the center of the resonance to access normal dispersion. For a red-detuning of 0.002 nm from the transmission peak of the resonance, the magnitude of dispersion is $- 320\textrm{ps}/\textrm{nm}$, which allows for 20 km of single mode fiber dispersion to be compensated for. At this wavelength, the incremental insertion loss is < 0.1 dB, which is a marginal increase in loss.
We further characterize the properties of the ring resonator when subject to thermo-optic tuning. Figure 3(a) shows the wavelength shift of the resonance from the drop port as a function of wavelength across the temperatures measured. The resonant wavelength increases with temperature, indicating a positive thermo-optic coefficient. The resonant peaks in the drop port are observed to red-shift by ∼0.34 nm for every 10°C increase in temperature which corresponds to a temperature dependent wavelength shift of 34pm/°C. It is further observed from Fig. 3(a) that the peak group delay remains relatively constant at 3300ps as temperature is varied between 20°C to 70°C.
Fig. 3. (a) Transmission spectrum (blue) and measured group delay (red) as a function of temperature (b) Measured dispersion at the drop-port near 1550 nm as a function of temperature. (c) Measured magnitude of dispersion (diamond) and dispersion slope (triangles) as a function of temperature. (d) Quality factor of the ring resonator as a function of temperature.
Figure 3(b) plots the measured dispersion spectrum for different temperatures. The magnitude of dispersion at a temperature of 20°C is experimentally measured to vary from to $- 12.3 \times {10^3}ps/nm$. It is observed from Fig. 3(a) and (b) that tuning the temperature from 20°C to 70°C results in a wavelength shift in the resonance and associated dispersion by 1.7 nm (∼200 GHz). Figure 2(d) further shows the dispersion of a single resonance as a function of temperature detuning, where it may be observed that a temperature change of ..0.5°C covers the entire dynamic range of dispersion $\; (\; + 12.9 \times {10^3}ps/nm$ to $- 12.3 \times {10^3}ps/nm$) available at the drop port. In other words, a large range of dispersion can be used for dispersion compensation using low power thermo-optic tuning. In Fig. 3(c), the peak-to-peak dispersion was measured, showing that the values remain relatively constant as temperature is increased. Insight into this phenomenon may be gleaned from Fig. 3(d), which plots the measured quality factor as a function of temperature. It is observed in Fig. 3(d) that the quality factor of the resonator remains relatively constant at $9 \times {10^4}$. Therefore, we expect the dispersion to be maintained at a constant magnitude in line with the trend of the quality factor as a function of temperature.
4. Dispersion compensation of high-speed data
Next, we perform high-speed data experiments, by using the ring resonator to compensate for the dispersion in 20 km single mode fiber (SMF) (320ps/nm). As shown in Fig. 2(c), the dispersion value to compensate for the dispersion through the 20 km SMF lies close to the resonant peak as there is a steep dispersion slope at the resonance of the drop port. Figure 4(a) shows the schematic of the high-speed experiments performed. A 231−1 Pseudo Random Binary Sequence (PRBS) non-return-to-zero (NRZ) 25 Gb/s signal was used to modulate a CW laser using an Optical Transmitter. The modulated optical signal was then sent through a 20 km SMF spool and the output of the 20 km SMF is degraded due to dispersion in the fiber. Thereafter, the optical signal is amplified using an erbium-doped fiber amplifier (EDFA) and filtered using an optical bandpass filter (OBPF) to eliminate the amplified spontaneous emission noise, before being coupled into the DUT (Si3N4 ring resonator device). The optical signal from the drop-port is then demodulated using a PIN-TIA photoreceiver. The Bit Error Rate Tester (BERT) and Digital Sampling Oscilloscope (DSO) was used to measure the bit error rates and eye diagram, respectively. To fix the wavelength/resonance from the DUT during the dispersion compensation experiments, the temperature was maintained at a fixed temperature using a thermo-electric temperature controller. The wavelength of the laser was then tuned to implement the required dispersion compensation magnitude of −320ps/nm.
Fig. 4. (a) Schematic for the high-speed measurement through the device under test (DUT) to compensate for dispersion. Captured eye-diagrams with 25Gb/s optical signal (b) for back-to-back characterization, (c) after propagation through 20 km SMF and (d) after dispersion compensation using the resonator drop port. (e) Bit Error Rate (BER) as a function of various data rates (Gb/s) with dispersion compensation (blue circle) and without dispersion compensation (red triangle).
Figure 4(b) shows the eye diagram for a back-to-back configuration measured from the DSO. Back-to-back here involves replacing the 20 km SMF and the DUT with their equivalent loss stage. Jitter was observed because the line rate was set to 25Gb/s which is near the upper end limit of the Optical Transmitter used. Figure 4(c) shows the eye diagram after the optical signal was transmitted through a 20 km SMF fiber without dispersion compensation with high degradation. Figure 4(d) shows the eye opening after the output from the 20 km fiber propagates through the Si3N4 ring resonator DUT. It is observed that there is a significant improvement in the eye diagram as compared to Fig. 4(c).
Next, we measured the Bit Error Rate (BER) using a BERT for the different available NRZ data rates (8.5, 10, 10.3125, 12.16512, 14.025, 25, 25.78125 Gb/s). The data rates chosen are close to data rates used in commercial SFP+ and SFP28 (Small Form Factor Pluggable) transceivers. These include companies in the parallel single mode 4 (PSM4) [14], course wavelength division multiplex 4 (CWDM4) [15] and 100G Serial Lambda [16] multi-source agreements. The measured bit error rates are plotted in Fig. 4(e). It is observed that an increasing data rate will increase the BER, consistent with dispersion impairments becoming more severe at higher data rates [17,18]. A high data rate implies a smaller bit duration. Assuming a finite jitter induced by chromatic dispersion from the fiber, higher data rates will experience greater jitter degradation. From 8.5Gb/s to 14.025 Gb/s, dispersion compensation improved the BER from ${10^{ - 4}}\; $ to the ${10^{ - 12}}\; $ region, with lowest BER of $4.15 \times {10^{ - 12}}\; $ using a data rate of 8.5Gb/s. The BER improved significantly from the ${10^{ - 3}}$ region to the ${10^{ - 11}}\; $ region for data rates of 25 and 25.78125 Gb/s. For this experiment, the only variable changed was the data rate, and the power received at the photoreceiver was fixed. The less sensitive PIN-TIA photoreceiver used in the setup would require greater optical power to be detected at higher data rates, and hence the BER was not improved to a near error free rate of ${10^{ - 12}}$. However, assuming that the Forward Error Correction threshold limit is $5 \times {10^{ - 5}}$, by implementing our Si3N4 ring resonator device, the dispersion compensation achieved with the Si3N4 ring resonator has restored the data to satisfy the minimum BER for transmission at rates up to 25.78125 Gb/s.
5. Discussion and conclusion
Table 1 shows the comparison of the figure of merits between this work and the state-of-the-art ring resonators used for dispersion compensation. We note that our work successfully demonstrates dispersion compensation of high-speed data impaired by dispersion in 20 km of optical fiber, at commercially adopted high-speed data rates (25 and 25.78125 Gb/s). Prior work as elucidated in Table 1 focused largely on simulations of improved bit error rates as well as lower data rates of up to 10 Gb/s. Our work experimentally showcases quantitative (bit-error rate) and qualitative (eye diagram) demonstrations of dispersion compensation of optical data using a ring resonator, thermally tuned to access the required magnitude of dispersion.
Table 1. Comparison of this work and state-of-the-art ring resonators
Another distinguishing point of this work is the utilization of the drop-port as compared to previous state-of-the-art devices which utilized the through port. In wavelength division multiplexed (WDM) systems, the utilization of the drop port for dispersion compensation is especially beneficial operationally, as compared to the use of the through port. Using the drop port configuration facilitates the simultaneous demultiplexing at each wavelength of light while providing a tailored magnitude of dispersion to compensate for the fiber dispersion that is unique to each WDM channel. If the ring resonators are used for post-compensation, they may be used for simultaneous dispersion compensation and de-multiplexing.
In practice, operational variations in the exact dispersion magnitude that needs to be compensated for can be addressed using thermo-optic tuning of the resonant wavelength. We note further that the dispersion magnitude achievable using the ring resonator is large, especially when compared with other alternatives. The options for dispersion compensation devices on a chip possess trade-offs which require careful consideration. These options include Mach-Zehnder interferometers, all-pass ring resonators [26] and chirped Bragg gratings [27–31]. Typically, devices which provide large operating bandwidths have smaller path length normalized dispersion. Gratings and Mach Zehnder Interferometer based dispersion compensation devices fall into this category. Also, the group delay in Bragg gratings may possess group delay ripple if apodization is not effectively implemented. This phenomenon is absent with ring resonators and a high group delay allows for compensation for larger time delays and in other words, allows for larger dispersion compensation. Conversely, devices which provide very large magnitudes of dispersion suffer from limited bandwidth. Such is the case for the resonator-based dispersion compensation we show here. Being that dispersion compensation demonstrated here relies on a resonant effect, the inherent operating bandwidth is small. However, the thermo-optic tuning demonstrated here allows a continuous, wide dynamic range of dispersion values to be accessed, while providing a lever to account for variations in both environment and manufacturing variations. To further optimize the performance of the resonator to accommodate higher data rates, resonators with lower quality factor can be utilized to have a larger bandwidth and this could be implemented easily if the length of fiber for which dispersion compensation is to be applied is not too long. Furthermore, the FSR of the resonator may also be tailored by changing the round-trip path length. This could be used as an additional design knob to tailor the dispersion.
We note further that one of the barriers to implementation of dispersion compensation at the chip-level is the achievable $dispersion\; magnitude/loss$. Dispersion impairments scale with fiber lengths and data rates. In an ideal scenario, integrating the dispersion compensation component within the transceiver unit would not introduce additional loss while allowing very large fiber lengths to be compensated for. In our work, there is only an incremental loss of 0.1 dB to operate at the resonant detuning to access dispersion compensation of 20 km of optical fiber through the drop port. The incremental insertion loss refers to the additional loss incurred by the optical signal when operating at the wavelength of interest for dispersion compensation compared to transmission at the center of the resonance. The ring resonator through port itself has a loss of less than 0.3 dB. Therefore, the loss experienced by the light is less than 0.4 dB. In addition, we note that the demonstrated 20 km of optical fiber exceeds the 10 km reach stipulated in the IEEE 802.3ba 100GBASE-LR4 standard. Cisco for example has recently reported transceivers designed for 10 km (LR4) reaches and their possible extension to 20 km reaches [17]. Consequently, the work here may provide a promising solution to dispersion impairments introduced by optical fibers, especially over very long reaches. The dispersion magnitudes which can be accessed in the demonstrated device are as large as from $\; + 12.9 \times {10^3}ps/nm$ to $- 12.3 \times {10^3}ps/nm$. To maintain a total loss of less than 3 dB with this device, the maximum dispersion magnitude which can be accessed is about $- 12.3 \times {10^3}ps/nm$ and this magnitude of dispersion matches that required for 760 km of optical fiber. In the future, the integration of micro-heaters could provide an additional pathway towards mitigating the limited device bandwidth or rate of thermal tuning, by allowing low-power, high-resolution control of the wavelength sensitive dispersion. The rate of which the device is being tuned thermally is intrinsically dependent on the thermal diffusivity of silicon nitride, which has a range of 0.32–0.84 $\; c{m^2}{s^{ - 1}}$ [32]. Since integrated micro-heaters provide localized heating to a targeted part of the chip (resonator), this would offer faster tuning than using an off-chip thermoelectric controller. Multiple dispersion compensating devices operating at different channel bands can be utilized to compensate for the limited device bandwidth. Within the device bandwidth from this device, a small change in temperature is only required to access a wide range of dispersion for compensation. The large range of normal or anomalous dispersion that can be accessed allows for compensation of longer fiber lengths, such as a 20 km SMF shown here. Greater distances in the hundreds would then be more dependent on the rate of thermal tuning that can be counteracted with delay lines or implementing a post-compensation scheme.
The demonstrated dispersion compensation device offers an on-chip method in which to manage dispersion impairments associated with the transmission of high-speed data in data center communications. The BER reduction from ${10^{ - 3}}$ to ${10^{ - 11}}\; $ for data rates of 25 and 25.78125 Gb/s represents a significant improvement brought about by the dispersion compensation provided by the ring resonator device. The low loss, high dispersion magnitude and thermo-optic tunability showcases the potential for performing integrated pre-compensation at the transceiver’s transmitter or post-compensation at the transceiver’s receiver using such devices. High-speed data experiments further demonstrate the dispersion compensation functions through the significant improvements in the eye diagram and bit error rates of non-return to zero data.
Funding
National Research Foundation Singapore (NRF-CRP18-2017-03); Ministry of Education - Singapore (ACRF Tier 2 Grant).
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. R. Nagarajan, I. Lyubomirsky, and O. Agazzi, “Low Power DSP-based Transceivers for Data Center Optical Fiber Communications,” J. Lightwave Technol. 39(16), 5221–5231 (2021). [CrossRef]
2. P. J. Winzer, D. T. Neilson, and A. R. Chraplyvy, “Fiber-optic transmission and networking: the previous 20 and the next 20 years,” Opt. Express 26(18), 24190–24239 (2018). [CrossRef]
3. G. P. Agrawal, “Fiber-Optic Communication Systems,” in Fiber-Optic Communication Systems, 4th ed. (John Wiley & Sons, 2010), pp. 345–406.
4. E. Agrell, M. Karlsson, A. R. Chraplyvy, D. J. Richardson, P. M. Krummrich, P. Winzer, K. Roberts, J. K. Fischer, S. J. Savory, B. J. Eggleton, M. Secondini, F. R. Kschischang, A. Lord, J. Prat, I. Tomkos, J. E. Bowers, S. Srinivasan, M. Brandt-Pearce, and N. Gisin, “Roadmap of optical communications,” J. Opt. 18(6), 063002 (2016). [CrossRef]
5. C. Peucheret, N. Hanik, R. Freund, L. Molle, and P. Jeppesen, “Optimization of pre- and post-dispersion compensation schemes for 10-Gbits/s NRZ links using standard and dispersion compensating fibers,” IEEE Photonics Technol. Lett. 12(8), 992–994 (2000). [CrossRef]
6. T. Xu, G. Jacobsen, S. Popov, J. Li, E. Vanin, K. Wang, A. T. Friberg, and Y. Zhang, “Chromatic dispersion compensation in coherent transmission system using digital filters,” Opt. Express 18(15), 16243–57 (2010). [CrossRef]
7. S. Echeverri-Chacon, J. J. Mohr, J. J. V. Olmos, P. Dawe, B. V. Pedersen, T. Franck, and T. Franck, “Transmitter and Dispersion Eye Closure Quaternary (TDECQ) and its sensitivity to impairments in PAM4 waveforms,” J. Lightwave Technol. 37(3), 852–860 (2019). [CrossRef]
8. H. Li, G. Balamurugan, M. Sakib, J. Sun, J. Driscoll, R. Kumar, H. Jayatilleka, H. Rong, J. Jaussi, and B. Casper, “A 112 Gb/s PAM4 Silicon Photonics Transmitter with Microring Modulator and CMOS Driver,” J. Lightwave Technol. 38(1), 131–138 (2020). [CrossRef]
9. J. Lambrech, J. Verbist, H. Ramon, M. Vanhoecke, J. Bauwelinck, X. Yin, and G. Roelkens, “Low-Power (1.5 pJ/b) Silicon Integrated 106 Gb/s PAM-4 Optical Transmitter,” J. Lightwave Technol. 38(2), 432–438 (2020). [CrossRef]
10. H. Li, G. Balamurugan, M. Sakib, R. Kumar, H. Jayatilleka, H. Rong, J. Jaussi, and B. Casper, “A 3D-Integrated Microring-Based 112Gb/s PAM-4 Silicon-Photonic Transmitter with Integrated Nonlinear Equalization and Thermal Control,” Dig. Tech. Pap. - IEEE Int. Solid-State Circuits Conf. (2020), 208–210.
11. M. V. Ramana Murty, J. Wang, A. L. Harren, A. Cheng, D. W. Dolfi, Z. Feng, A. Sridhara, S. T. Joyo, J. Chu, and L. M. Giovane, “Development and Characterization of 100 Gb/s Data Communication VCSELs,” IEEE Photonics Technol. Lett. 33(16), 812–815 (2021). [CrossRef]
12. G. Agrawal, “Group-Velocity Dispersion,” in Nonlinear Fiber Optics (Elsevier, 2013), pp. 57–85.
13. B. E. A. Saleh and M. C. Teich, “Fiber-Optic Communications,” Fundamentals of Photonics (Wiley-Interscience, 2003), pp. 874–917.
14. PSM4 MSA Group, “100G PSM4 MSA Specification,” PSM4 (2014), http://www.psm4.org/100G-PSM4-Specification-2.0.pdf.
15. M. Fields, “Objective of the CWDM4 MSA,” CWDM4 MSA, http://www.cwdm4-msa.org/files/CWDM4_MSA_4-29-14_FINAL.pdf.
16. L. Msa, “100G-LR1-20, 100G-ER1-30, 100G-ER1-40 Technical Specifications,” 100G Lambda MSA (2021), https://100glambda.com/specifications.
17. M. Traverso, M. Mazzini, K. Lukshmikumar, S. Sunder, A. Kurylak, C. Appel, C. Muzio, R. Tummidi, A. Cervasio, M. Nadeau, W. Li, J. Nelman, and M. Webster, “Integrated Silicon Photonics Transceiver Module for 100Gbit/s 20 km Transmission,” in 2021 Optical Fiber Communications Conference and Exhibition, OFC 2021 - Proceedings (2021), vol. 1, 20.
18. R. Nagarajan, M. Filer, Y. Fu, M. Kato, T. Rope, and J. Stewart, “Silicon photonics-based 100 Gbit/s, PAM4, DWDM data center interconnects,” J. Opt. Commun. Netw. 10(7), B25–B36 (2018). [CrossRef]
19. C. K. Madsen, G. Lenz, A. J. Bruce, M. A. Cappuzzo, L. T. Gomez, and R. E. Scotti, “Integrated All-Pass Filters for Tunable Dispersion and Dispersion Slope Compensation,” IEEE Photonics Technol. Lett. 11(12), 1623–1625 (1999). [CrossRef]
20. C. K. Madsen, G. Lenz, A. J. Bruce, M. A. Capuzzo, L. T. Gomez, T. N. Nielsen, and I. Brener, “Multistage dispersion compensator using ring resonators,” Opt. Lett. 24(22), 1555 (1999). [CrossRef]
21. H. Takahashi, R. Inohara, K. Nishimura, and M. Usami, “Expansion of Bandwidth of Tunable Dispersion Compensator Based on Ring Resonators Utilizing Negative Group Delay,” J. Lightwave Technol. 24(6), 2276–2286 (2006). [CrossRef]
22. J. Montalvo and C. Vázquez, “Ring resonator with an internal Sagnac loop for dispersion compensation in DWDM backbone networks,” Proc. SPIE 6593, 65931E (2007). [CrossRef]
23. C. Ferrari, F. Morichetti, L. Flaccadori, and A. Melloni, “Chromatic dispersion compensation in a ring resonator wavelength router,” in Proc. 2008 10th Anniv. Int. Conf. Transparent Opt. Networks, Ict. (2008), pp. 199–202.
24. J. Riemensberger, K. Hartinger, T. Herr, V. Brasch, R. Holzwarth, and T. J. Kippenberg, “Dispersion engineering of thick high-Q silicon nitride ring-resonators via atomic layer deposition,” Opt. Express 20(25), 27661–27669 (2012). [CrossRef]
25. S. Kim, K. Han, C. Wang, J. A. Jaramillo-Villegas, X. Xue, C. Bao, D. E. Leaird, A. M. Weiner, and M. Qi, “Dispersion engineering and frequency comb generation in thin silicon nitride concentric microresonators,” Nat. Commun. 8(1), 372 (2017). [CrossRef]
26. C. K. Madsen and G. Lenz, “Optical All-Pass Filters for Phase Response Design with Applications for Dispersion Compensation,” IEEE Photonics Technol. Lett. 10(7), 994–996 (1998). [CrossRef]
27. J. W. Choi, E. Sahin, B. U. Sohn, G. F. R. Chen, D. K. T. Ng, A. M. Agarwal, L. C. Kimerling, and D. T. H. Tan, “High spectro-temporal compression on a nonlinear CMOS-chip,” Light: Sci. Appl. 10(1), 130 (2021). [CrossRef]
28. E. Sahin, K. J. A. Ooi, C. E. Png, and D. T. H. Tan, “Large, scalable dispersion engineering using cladding-modulated Bragg gratings on a silicon chip,” Appl. Phys. Lett. 110, 161113 (2017). [CrossRef]
29. G. F. R. Chen, T. Wang, C. Donnelly, and D. T. H. Tan, “Second and third order optical dispersion using nonlinearly chirped silicon waveguide gratings,” Opt. Express 21(24), 29223–29230 (2013). [CrossRef]
30. C. S. Goh, S. Y. Set, and K. Kikuchi, “Design and Fabrication of a Tunable Dispersion-Slope Compensating Module Based on Strain-Chirped Fiber Bragg Gratings,” IEEE Photonics Technol. Lett. 16(2), 524–526 (2004). [CrossRef]
31. Y. J. Lee, J. Bae, K. Lee, J. M. Jeong, and S. B. Lee, “Tunable Dispersion and Dispersion Slope Compensator Using Strain-Chirped Fiber Bragg Grating,” IEEE Photonics Technol. Lett. 19(10), 762–764 (2007). [CrossRef]
32. B. Li, L. Pottier, J. P. Roger, D. Fournier, K. Watari, and K. Hirao, “Measuring the anisotropic thermal diffusivity of silicon nitride grains by thermoreflectance microscopy,” J. Eur. Ceram. Soc. 19(8), 1631–1639 (1999). [CrossRef]
