We characterize the interferometric crosstalk and system performance of two optical add-drop multiplexer (OADM) designs based on Bragg grating/Mach-Zehnder interferometers implemented in silicon-on-insulator. Both OADM designs exhibit low crosstalk and negligible crosstalk-induced power penalties over their 3 dB bandwidths. The devices are tolerant to wavelength drift and misalignment between the transmitter and OADM; moreover, their designs can be optimized further to enable high performance operation in WDM systems.
© 2015 Optical Society of America
WDM is a promising solution to meet the demand of increasing bandwidth (BW) in fiber optic communications and short reach optical interconnections [1,2]. Optical add-drop multiplexers (OADMs) are essential to provide wavelength selectivity and to route different WDM channels; they have been implemented using a variety of filters such as thin-film , liquid crystal , microring resonators , and Bragg gratings (BGs) . From an integration perspective, ring resonators and BGs are attractive and a number of device configurations have been reported . Ring resonators are intrinsically periodic filters and a large free spectral range (FSR) is more challenging to achieve as this requires a very small ring radius; moreover, cascaded rings are required in order to tailor the spectral response. Thus, they may not be suitable for single channel or non-periodic (e.g., two consecutive channels) add-drop operations. On the other hand, the use of BGs in OADMs may overcome such issues. Traditionally, a significant drawback of BG-based OADMs is the need for optical circulators, unless the gratings are incorporated in coupling or interferometric structures [8–18]. In particular, transmission bandpass filters and OADMs incorporating BGs in Michelson interferometers, Sagnac interferometers, and Mach-Zehnder interferometers (MZIs) have been demonstrated in both fiber and planar waveguide implementations [8–17]. During the past few years, there has been a growing interest in silicon-on-insulator (SOI), motivated by the requirements of and their application in chip-level communications and to exploit CMOS compatibility . The development of OADMs in SOI can lead to further integration of WDM subsystems and several approaches have been reported, such as grating-assisted contra-directional couplers, which tend to be complex and require careful adjustment of waveguide widths [20,21], and BG/MZI configurations , in which a very limited transmission isolation of only 8 dB was achieved. One key characteristic of OADMs is interferometric (intrachannel) crosstalk which can severely degrade system performance. Interferometric crosstalk occurs when the drop and add channels use the same wavelength and the OADM does not have infinite isolation, i.e., the BG is not 100% reflective. While the impact of interferometric crosstalk in fiber and silica-based BG/MZI OADMs has been investigated [13,14,17], similar studies have not been performed for SOI-based implementations. Moreover, there have been no demonstrations of SOI-based BG/MZI OADMs with high transmission isolation. Such studies and demonstrations are needed to assess the suitability of the SOI-based BG/MZI OADMs as a viable technology.
In this paper, we report BG/MZI OADMs in SOI with a much higher transmission isolation of > 40 dB compared to 8 dB demonstrated in ; moreover, we implement a configuration based on cascaded BG/MZI OADMs to reduce further interferometric crosstalk . We characterize and compare the interferometric crosstalk performance for both designs, including bit error rate (BER) measurements. The OADMs have error-free operation with negligible crosstalk-induced power penalties over the operating channel 3 dB bandwidth (BW). The SOI-based implementations also have a compact footprint of only 391 μm × 1.4 mm or 391 μm × 3.6 mm. While the BW of the OADMs (~3.5 nm) enable simultaneous multi-channel (e.g., two consecutive channels) add-drop, they are not suitable for single channel operation with narrow channel spacing. However, the BGs can be designed and optimized to support narrower channel spacing while maintaining a compact footprint.
2. BG/MZI OADM designs
Figure 1 shows a schematic of the two OADM designs, both of which are four port devices with input, drop, add, and through ports. The conventional BG/MZI OADM, denoted Design #1 and shown in Fig. 1(a), is a design with a well understood principle of operation . It comprises a pair of (identical) BGs resonant at λB that are placed symmetrically within the interferometer, i.e., L1 = L2 and L3 = L4 (and correspondingly, L1 + L3 = L2 + L4). Coupling of the MZI arms is through multimode interference (MMI) couplers, each with a splitting ratio of 0.5. Light launched into the input port is split evenly by MMI1 and propagates via the two paths (L1 and L2) to the BGs. Light at λB is reflected back to MMI1 and exits through the drop port while the remaining wavelengths propagate to MMI2 and exit via the through port. Since the device is symmetric, light at λB launched into the add port can be extracted at the through port.
Figure 1(b) shows a schematic of the cascaded BG/MZI OADM, denoted Design #2 . The add and drop ports operate in the same manner as in the conventional BG/MZI OADM. The additional intermediate BG, which connects the two BG/MZI structures, intercepts potential leakage, i.e., if an input signal at λB is launched into the input port and the reflectivities of the BGs in the BG/MZI structures are not sufficiently high, the intermediate BG further reduces the leakage of light from propagating to the through port of the device. Similarly, the intermediate BG rejects further the residual signal at λB from appearing at the drop port when a signal is launched in the add port.
The cross-section of the silicon waveguides used in the devices is shown in Fig. 1(c). The width and height are 500 nm and 220 nm, respectively; the waveguides sit on top of a 3 µm thick buried oxide (BOX) layer and are covered by a 2 µm thick index-matched top oxide cladding. The (uniform) BGs are based on sidewall corrugations  [see Fig. 1(d)] with a corrugation depth of nm. Each grating has a period Λ = 320 nm and comprises 3,000 periods, corresponding to a grating length of 960 μm. The input and output waveguides of the MMI coupler are tapered from 500 nm to 1.5 µm over a length of 20 µm. The MM waveguide in the coupler has a width of 6 µm and a length of 128 µm in order to achieve a 50:50 splitting ratio. The two arms in the interferometer are separated by only 23 µm (using S-bend waveguides with a 5 µm bend radius) in order to reduce mismatch of the BGs during fabrication. Light is coupled in and out of the device via vertical grating couplers (VGCs)  that are optimized for TE transmission. The center-to-center spacing between consecutive VGCs is 127 μm (each VGC occupies 25 μm × 35 μm) and they are aligned on the same side of the chip to facilitate coupling with a fiber ribbon array. The typical coupling loss of an input and output VGC pair (connected by a short length of waveguide) is 16 dB and can be reduced significantly using an improved coupling setup and VGC design . The devices were fabricated at the University of Washington Nanofabrication Facility using electron beam lithography with a single full etch.
3. Characterization of the BG/MZI OADMs
Figure 2 shows the experimental setup for characterizing the BG/MZI OADMs. We use a four channel fiber ribbon array to enable simultaneous launching of light into the input and add ports and collection of the optical signals from the drop and through ports. Each transmitter comprises a tunable external cavity laser (ECL), a polarization controller (PC), an electro-optic Mach-Zehnder modulator (EO-MZM) driven by its own data sequence from a pulse pattern generator (PPG), and an erbium-doped fiber amplifier (EDFA). For detection, we use either a power meter, an optical spectrum analyzer, or a pre-amplified receiver. The BG/MZI OADMs are characterized in terms of their spectral responses, crosstalk, and BER performance.
3.1. Spectral measurements
To measure the spectral responses of the two BG/MZI OADMs, we operate the transmitters in CW mode and use a power meter to detect the output signal as the ECLs are scanned in wavelength. All of the measured spectra are normalized to the nominal transfer function of the VGCs, which is characterized separately via a short length of waveguide. For Design #1, the insertion losses (above the VGC losses) at the Bragg wavelength from the input port to drop port and from the add port to through port are ~6 dB and 7 dB, respectively, while that from the input port to through port is ~7 dB. For Design #2, the insertion losses at the Bragg wavelength from the input port to drop port and from the add port to through port are ~5 dB and 6 dB, respectively. On the other hand, the insertion loss from the input port to through port increases to ~13 dB due to propagation losses through the additional components and waveguide lengths. Figure 3(a) shows the measured spectral responses of the drop, add, and through ports of Design #1. The drop and the add responses both have passbands centered at 1559.1 nm with a 3 dB BW of 3.4 nm. Although the device is symmetric, the drop and add responses do not have the exact same spectral response, which is due in part to the non-identical spectral responses of the BGs, VGCs, and MMI couplers, as well as the difference in waveguide paths due to fabrication errors, i.e., a slight difference between L1 and L2 or L3 and L4 will create a slowly-varying sinusoidal spectral response (from the misbalanced interferometer) that modulates the BG spectral response. The input to through response shows a transmission isolation of more than 40 dB at the Bragg wavelength. Figure 3(b) shows the measured responses of the drop, add, and through ports of Design #2. The drop and add responses both have passbands centered at 1558.2 nm with a 3 dB BW of 3.5 nm. Even though both OADM designs employ the same grating parameters, the differences in BW and Bragg wavelength are due to variations in fabrication. The input to through response exhibits a transmission isolation of more than 45 dB at the Bragg wavelength. The higher transmission isolation for Design #2 is due to the cascaded structure incorporating the intermediate BG. Note, however, that the improvement in performance for the cascaded structure will be limited by any misbalance in the interferometer and/or non-identical characteristics of all the BGs as these will contribute to reducing the effective reflectivity of the BG and hence transmission isolation.
Since we use uniform BGs, the spectral responses of the OADMs exhibit relatively high sidelobes. These can be reduced using apodized grating designs as demonstrated in . To reduce the 3 dB BW, longer BGs are required. Owing to the large index contrast in SOI, it is possible to use spiral waveguide BGs; indeed, this approach has been used to realize cm long BGs with high transmission isolation (40 dB), 3 dB BWs ~0.5 nm, and that occupy a total footprint of only 189 μm × 189 μm [27,28].
Our two OADM designs have a transmission isolation that is generally 10 dB higher than that of FBG/MZI OADMs [13,14]; however, they are at least 5 dB less than that of the silica-based BG/MZI OADMs reported in . With proper design, it may be possible to increase the transmission isolation of the SOI BG/MZI OADM further using longer grating structures, without increasing significantly the footprint (e.g., with waveguide spirals).
3.2. Interferometric crosstalk
To evaluate the interferometric crosstalk, we launch a drop signal (CW) and an add signal (CW) at the same wavelength within the grating BW (e.g., at 1558.1 nm) and both with the same power at the input and add ports, respectively. We compare the spectral output at the drop port when the signal is launched at the input or add port only, giving the interferometric crosstalk for the drop operation. Similarly, we compare the spectral output at the through port when the signal is launched at the add or input port only, giving the interferometric crosstalk for the add operation. Figures 4(a) and 4(b) show the spectral output at the drop and add ports of Design #1 for the two different cases. The values of interferometric crosstalk are −49.21 dB and −47.98 dB for the drop and add operations, respectively. In principle, since the device is symmetric, the interferometric cross-talk levels should be the same; however, as explained in Section 3.1, there are several factors which can contribute to creating an asymmetric response. Figures 4(c) and 4(d) show the spectral output at the drop and add ports for Design #2 for the two different cases. The values of interferometric crosstalk are −52.33 dB and −50.49 dB for the drop and add operations, respectively, which is a 4 dB improvement compared to Design #1.
The interferometric crosstalk (X) is directly related to the reflectivity (R) of the BGs as follows :13, 14]. The 4 dB reduction in interferometric crosstalk for Design #2 is less than that obtained for the cascaded FBG/MZI configuration (see, e.g., ) and is due to (1) the fact that the reflectivities of the SOI BGs are already quite high and the impact of cascading the structures to enhance further the effective reflectivity at the Bragg wavelength is less and (2) fabrication errors as described above.
We then measure the interferometric crosstalk over the grating reflection BW. Figures 5(a) and 5(b) compare the crosstalk for both BG/MZI designs for the drop and add operations, respectively (the x-axis denotes wavelength detuning from the center wavelength λB normalized to the 3 dB reflection BW). The mean values of interferometric crosstalk are −50.3 dB and −50.9 dB with corresponding standard deviations of 2.0 dB and 1.9 dB for the drop operation for Designs #1 and 2, respectively. The mean values of interferometric crosstalk are −50.5 dB and −54.7 dB with corresponding standard deviations of 2.3 dB and 2.1 dB for the add operation for Designs #1 and 2, respectively. The values of interferometric crosstalk are relatively flat over the entire 3 dB bandwidths of the OADMs, which result in very high utilization factors of 0.84 and 0.95 (defined as the ratio of the BW at an interferometric crosstalk level of −35 dB to the 20 dB BW of the main grating reflection response). It should be noted that due to the high reflectivity of the BGs, Design #1 has a similar level of crosstalk as with the FBG-based version of Design #2 .
3.3. BER measurements
We further assess the performance of the two OADMs using BER measurements. In these measurements, the transmitters are configured to generate independent 10 Gb/s NRZ-OOK signals with a pseudo random bit sequence (PRBS).
First, Fig. 6 shows the measured BER at the through port for an out-of-band signal at a wavelength of 1555 nm for both designs; corresponding eye diagrams are also shown. Both designs provide error-free operation with a power penalty less than 1 dB.
Second, we assess the interchannel crosstalk at the drop port. In this case, the two transmitters, one operating at the Bragg wavelength λB (i.e., the drop channel) and the second detuned from the main grating reflection response (i.e., the non-drop channel), are combined and launched into the input port with equal power. Due to the relatively high sidelobes associated with the use of uniform BGs, the rejection of the non-drop channel is as low as only 10 dB. Nevertheless, the worst-case power penalty is less than 1 dB (as expected, the power penalty is higher when the rejection of the non-drop channel is lower). Figure 7 summarizes the results for both designs.
Finally, to assess the interferometric crosstalk, the transmitters are operated at the same wavelength with one being launched into the input port and the other into the add port simultaneously. We then measure the BER of the drop channel with and without the interfering (add) signal and of the add channel with and without the interfering (drop) signal. In the absence of crosstalk, the power penalties for the drop channel compared to the back-to-back case are ~1.5 dB and 2.0 dB for Designs #1 and 2, respectively; the corresponding penalties for the add channel are 1.6 dB and 1.2 dB. Figures 7(b) and 8(a) summarize the crosstalk-induced power penalties for the drop and add operations, respectively. For the drop operation, both OADMs have negligible crosstalk-induced power penalty and have similar performance. For the add operation, the use of Design #1 incurs a penalty up to 0.6 dB higher compared to Design #2 (this agrees generally with the results shown in Fig. 5(b) where Design #2 has lower interferometric crosstalk at the add port). Note that for a given design, the penalties for the drop and add operations are not exactly the same due largely to the asymmetric response of the devices (differences in coupling losses may also have an impact since the add response uses the ports on the two outer edges of the fiber array while the drop response is measured using the two ports near the center of the fiber array). The measured interferometric crosstalk and crosstalk induced power penalties are relatively constant over the entire 3 dB BW of the SOI OADMs; as such, they are tolerant to wavelength drift and misalignment between the transmitter and OADM.
4. Discussion and summary
We have characterized and compared the performance of two BG/MZI OADMs in SOI. Conventional and cascaded (crosstalk-reduced) designs are examined; they exhibit BWs of ~3.4 nm and 3.5 nm at a wavelength of 1559.1 nm and 1558.2 nm, and transmission isolations of ~40 dB and at least 45 dB, respectively. We have investigated the crosstalk performance for both device designs and the crosstalk is low within the 3 dB BW of the OADM. 10 Gb/s BER measurements are used to verify the crosstalk performance of the OADMs. The interferometric crosstalk-induced power penalties at the drop port are negligible for both designs; at the add port, the conventional design has up to 0.6 dB higher power penalty compared to the cascaded design. Differences in the drop and add responses are observed and attributed to differences in the MMI couplers, propagation losses, and the spectral responses of the BGs and VGCs. By taking advantage of the high reflectivities of BGs in SOI, we have shown that the conventional SOI-based BG/MZI OADMs has comparable crosstalk performance compared to the cascaded design implemented with FBGs; moreover, the SOI devices have the advantage of a compact footprint. The SOI BGs can be optimized to (1) reduce sidelobe levels and interchannel crosstalk using apodization and (2) reduce operating BW (e.g., to 0.5 nm or below) to support single channel add-drop operation with narrower WDM channel spacing (e.g., 100 GHz) as well as transmission isolation by increasing the length of the grating structures using spiral waveguides without necessarily increasing the device footprint. Finally, all the structures are fabricated using a single etch, which eases the fabrication process. We believe that the OADMs presented can serve as fundamental components for more complex multi-channel add/drop multiplexing devices.
Devices were fabricated by Richard Bojko at the University of Washington Nanofabrication Facility (WNF), a member of the NSF National Nanotechnology Infrastructure Network. This research was supported in part by the Natural Sciences and Engineering Research Council of Canada via the CREATE NGON and Si-EPIC programs. We thank Paul Morin (McGill) for discussions.
References and links
1. A. Alduino and M. Paniccia, “Interconnects - wiring electronics with light,” Nat. Photonics 1(3), 153–155 (2007). [CrossRef]
2. C. Kachris and I. Tomkos, “A survey on optical interconnects for data centers,” IEEE Comm. Surv. and Tutor. 14(4), 1021–1036 (2012). [CrossRef]
3. K. Kintaka, J. Nishii, S. Murata, and S. Ura, “Eight-channel WDM intraboard optical interconnect device by integration of add/drop multiplexers in thin-film waveguide,” J. Lightwave Technol. 28(9), 1398–1403 (2010). [CrossRef]
4. N. A. Riza and S. F. Yuan, “Reconfigurable wavelength add-drop filtering based on a Banyan network topology and ferroelectric liquid crystal fiber-optic switches,” J. Lightwave Technol. 17(9), 1575–1584 (1999). [CrossRef]
5. B. E. Little, S. T. Chu, H. A. Haus, J. Foresi, and J. P. Laine, “Microring resonator channel dropping filters,” J. Lightwave Technol. 15(6), 998–1005 (1997). [CrossRef]
6. C. R. Giles, “Lightwave applications of fiber Bragg gratings,” J. Lightwave Technol. 15(8), 1391–1404 (1997). [CrossRef]
7. H. Subbaraman, X. Xu, A. Hosseini, X. Zhang, Y. Zhang, D. Kwong, and R. T. Chen, “Recent advances in silicon-based passive and active optical interconnects,” Opt. Express 23(3), 2487–2510 (2015). [CrossRef] [PubMed]
8. F. Bilodeau, K. O. Hill, B. Malo, D. C. Johnson, and J. Albert, “High-return-loss narrow-band all-fiber bandpass Bragg transmission Filter,” IEEE Photonics Technol. Lett. 6(1), 80–82 (1994). [CrossRef]
9. F. Bilodeau, D. C. Johnson, S. Theriault, B. Malo, J. Albert, and K. O. Hill, “An all-fiber dense-wavelength-division multiplexer demultiplexer using photoimprinted Bragg gratings,” IEEE Photonics Technol. Lett. 7(4), 388–390 (1995). [CrossRef]
10. G. E. Kohnke, C. H. Henry, E. J. Laskowski, M. A. Cappuzzo, T. A. Strasser, and A. E. White, “Silica based Mach-Zehnder add-drop filter fabricated with UV induced gratings,” Electron. Lett. 32(17), 1579–1580 (1996). [CrossRef]
11. Y. Hibino, T. Kitagawa, K. O. Hill, F. Bilodeau, B. Malo, J. Albert, and D. C. Johnson, “Wavelength division multiplexer with photoinduced Bragg gratings fabricated in a planar-lightwave-circuit-type a symmetric Mach-Zehnder interferometer on Si,” IEEE Photonics Technol. Lett. 8(1), 84–86 (1996). [CrossRef]
12. J. M. Jouanno, D. Zauner, and M. Kristensen, “Low crosstalk planar optical add-drop multiplexer fabricated with UV-induced Bragg gratings,” Electron. Lett. 33(25), 2120–2121 (1997). [CrossRef]
13. T. Mizuochi, T. Kitayama, K. Shimizu, and K. Ito, “Interferometric crosstalk-free optical add/drop multiplexer using Mach-Zehnder-based fiber gratings,” J. Lightwave Technol. 16(2), 265–276 (1998). [CrossRef]
14. R. J. S. Pedersen and B. F. Jorgensen, “Impact of coherent crosstalk on usable bandwidth of a grating-MZI based OADM,” IEEE Photonics Technol. Lett. 10(4), 558–560 (1998). [CrossRef]
15. L. Eldada, S. Yin, C. Poga, C. Glass, R. Blomquist, and R. A. Norwood, “Integrated multichannel OADM’s using polymer Bragg grating MZI’s,” IEEE Photonics Technol. Lett. 10(10), 1416–1418 (1998). [CrossRef]
16. J. Albert, F. Bilodeau, D. C. Johnson, K. O. Hill, K. Hattori, T. Kitagawa, Y. Hibino, and M. Abe, “Low-loss planar lightwave circuit OADM with high isolation and no polarization dependence,” IEEE Photonics Technol. Lett. 11(3), 346–348 (1999). [CrossRef]
17. K. Hattori, M. Abe, J. Albert, F. Bilodeau, K. O. Hill, Y. Hibino, T. Kitagawa, and K. Oguchi, “Coherent crosstalk of an optical add/drop filter with Bragg gratings in a PLC Mach-Zehnder interferometer for optical LAN,” IEEE Photonics Technol. Lett. 11(2), 272–274 (1999). [CrossRef]
18. D. Mechin, P. Grosso, and D. Bosc, “Add-drop multiplexer with UV-Written Bragg gratings and directional coupler in SiO2-Si integrated waveguides,” J. Lightwave Technol. 19(9), 1282–1286 (2001). [CrossRef]
19. A. Rickman, “The commercialization of silicon photonics,” Nat. Photonics 8(8), 579–582 (2014). [CrossRef]
21. H. Qiu, G. Jiang, T. Hu, H. Shao, P. Yu, J. Yang, and X. Jiang, “FSR-free add-drop filter based on silicon grating-assisted contradirectional couplers,” Opt. Lett. 38(1), 1–3 (2013). [CrossRef] [PubMed]
22. H. Yamada, T. Chu, S. Ishida, and Y. Arakawa, “Optical add-drop multiplexers based on Si-wire waveguides,” Appl. Phys. Lett. 86(19), 191107 (2005). [CrossRef]
23. X. Wang, W. Shi, H. Yun, S. Grist, N. A. F. Jaeger, and L. Chrostowski, “Narrow-band waveguide Bragg gratings on SOI wafers with CMOS-compatible fabrication process,” Opt. Express 20(14), 15547–15558 (2012). [CrossRef] [PubMed]
24. Y. Wang, X. Wang, J. Flueckiger, H. Yun, W. Shi, R. Bojko, N. A. F. Jaeger, and L. Chrostowski, “Focusing sub-wavelength grating couplers with low back reflections for rapid prototyping of silicon photonic circuits,” Opt. Express 22(17), 20652–20662 (2014). [CrossRef] [PubMed]
26. A. D. Simard, N. Belhadj, Y. Painchaud, and S. LaRochelle, “Apodized silicon-on-insulator Bragg gratings,” IEEE Photonics Technol. Lett. 24(12), 1033–1035 (2012). [CrossRef]
28. Z. Chen, J. Flueckiger, X. Wang, H. Yun, Y. Wang, Z. Lu, F. Zhang, N. A. F. Jaeger, and L. Chrostowski, “Bragg grating spiral strip waveguide filters for TM modes,” in Conference on Lasers and Electrop-Optics, OSA Technical Digest (OSA, 2015), paper SM3I.7. [CrossRef]