Open Access
Add to BibSonomy
Abstract
Silicon-photonic integrated circuits are a pivotal technology for the continued growth of data communications. A main task of silicon photonics is the wavelength division multiplexing of communication channels to aggregate bandwidths that exceed the working rates available in electronics. In this work, we design and implement a 16-channel, wavelength division multiplexing device in silicon-on-insulator. The device operates at the O-band wavelengths, centered at 1310 nm, which are favored by many data center applications. The spacing between adjacent channels is 0.96 nm (167 GHz), close to those of dense wavelength division multiplexing standards in the 1550 nm wavelength range (C band). The layout consists of 15 Mach-Zehnder interferometers, cascaded in a four-stage tree topology. The differential phase delay within each interferometer is precisely trimmed post-fabrication, through local illumination of a photosensitive upper cladding layer of As2Se3 chalcogenide glass. Trimming is performed subject to closed-loop feedback of transfer functions measurements. The devices can be useful in data center optical communications.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
The integration of multiple photonic devices at the chip level is essential for the continued growth of data communications. Silicon-on-insulator (SOI) is often the material platform of choice for photonic circuits, due to the mature fabrication technology and the promise of co-integration alongside electronic components [1,2]. The aggregate bandwidths of optical communications are too broad to be processed directly in the electrical domain. Therefore, a primary task of silicon photonics is the wavelength division multiplexing (WDM) of communication channels, each operating at bandwidths that are compatible with electronics. Metrics of interest of WDM devices include the number of channels supported, the spacing between channels, the extent of crosstalk among output ports, the spectral uniformity of passbands, and the sharpness of spectral transitions between pass and stop bands [3]. Many devices require the post-fabrication adjustments of transfer function parameters, using continuous active control or one-time trimming [4–7]. The complexity of adjustment protocols increases with the number of wavelength channels.
Numerous works have reported WDM devices in SOI, using arrayed waveguide gratings, echelle gratings, cascaded Mach-Zehnder interferometers, or ring resonator layouts [8–19]. Many devices were built for operation in the C-band wavelength range of 1530-1565 nm [8–19]. In recent years, several circuits were designed for the O band, centered at 1310 nm wavelength, which is favored in many data center applications due to the availability of lower cost light sources [20–30]. In one example, Davis et al. demonstrated a 16-channel WDM device in the O band based on concatenated Bragg grating filters, with channels spacing of 2.6 nm [20]. Very recently, Akiyama and coworkers demonstrated a 32-channel device with impressive performance of 50 GHz channels spacing and crosstalk levels below −38 dB [30]. The layout of that state-of-the-art device is comprised of 93 Mach-Zehnder interferometers [30].
In this work, we report a 16-channel, O-band WDM device in the standard SOI platform. The device is comprised of 15 Mach-Zehnder interferometers cascaded in a four-stage tree topology [31]. Differential phase delays within each interferometer stage are trimmed post-fabrication, using one-time illumination of a photo-sensitive upper cladding layer [32–37]. The spacing between adjacent channels is 0.96 nm (167 GHz). The devices can be useful for photonic integrated data communications.
2. Principle of operation
A schematic illustration of the 16-channel WDM device realized in this work is shown in Fig. 1. Light from a common input port is split into 16 outputs using four stages of cascaded imbalanced Mach-Zehnder interferometers in a tree topology [31,37]. The entire layout consists of 15 interferometers. The power splitting ratios of the directional couplers at the input and output of all interferometers are 50/50. The differential group delay of the first interferometer at the input (“A” stage) equals $8\tau $, where $\tau $ is a delay unit. The subsequent “B”, “C” and “D” stages consist of two, four and eight interferometers in parallel, with respective differential group delays of $4\tau $, $ 2\tau $ and $\tau $ (Fig. 1).
Fig. 1. Schematic illustration of a 16-channel, wavelength division multiplexer device, comprised of 15 Mach-Zehnder interferometers in a four-stage, cascaded tree topology. Differential group delays and differential phases within each interferometer are noted. The orange-colored patches represent the locations of deposited regions of a photosensitive upper cladding layer, used in post-fabrication trimming of the differential phase delay within each interferometer ( [34–37], see below). TP: test port. The 16 outputs ports are on the right side.
The waveform at each of the 16 output ports represents the sum of 16 delayed replicas of the common input. The response of each output $m = 1 \ldots 16$ can be represented in the $Z$-domain in terms of 15 zero locations [31,37]:
All zeros ${z_{m,l}}$ of all response functions are located on the unit circle. Their specific locations are determined by the exact differential phases within each interferometer stage at a reference optical frequency ${\omega _0}$. The designed phases are detailed in Table 1 [31]. With these choices of differential phases, the angles of all zeros’ locations become integer multiples of $2\pi /16$ [31]. The responses of the 16 output ports differ as follows: ${H_m}(z )$ includes zeros at all locations $2\pi k/16$, $k = 1 \ldots 16$ and $k \ne m$, however no zero appears at $2\pi m/16$ (see Fig. 2(a)). The frequency response of output port m is therefore characterized by a passband centered at a normalized frequency ${\mathrm{\nu }_m} = m/16$, $0 \le {\mathrm{\nu }_m} \le 1$ (Fig. 2(b)).
Fig. 2. (a): Locations of zeros in the $Z$-domain response ${H_6}(z )$ of output port $m = 6$. 15 zeros appear on the unit circle, at all integer multiples $2\pi k/16$, $k = 1 \ldots 16$, except for $k = m = 6$ where no zero exists. Blue markers show the locations of 8 zeros corresponding to the stage A Mach-Zehnder interferometer, red markers indicate the four zeros of the B stage, the two green markers represent zero locations of the C stage, and the magenta marker corresponds to the D stage. (b): Normalized transfer function of optical power between the common input and output port $m$ = 8 as a function of normalized frequency.
Table 1. Designed differential phase delays within the 15 Mach-Zehnder interferometer stages of a 16-channel, wavelength division multiplexer device. (For the labeling of differential phase variables, see Fig. 1).
The proper function of the WDM device depends critically on the correct values of differential phase delays as specified in Table 1. The phases cannot be controlled in open-loop fabrication with sufficient precision due to inevitable residual uncertainties in the widths and depths of etched waveguide patterns. The differential phases therefore require post-fabrication trimming. Possible solutions include the application of current to local heating elements [4,5], and the injection or depletion of carriers to p-n junction regions formed across waveguides [6,7]. In this work, we employ one-time, post-fabrication trimming of passive devices through the local illumination of a photo-sensitive upper cladding layer ( [34–37], see next section).
3. Device fabrication and testing
Devices were fabricated by the commercial silicon foundry Tower Semiconductors in Migdal Ha’Emek, Israel [37]. Standard 8” SOI wafers were used. The thicknesses of the silicon device layer and the buried oxide layer were 220 nm and 2 µm, respectively. Ridge waveguides were defined in the silicon device layer using stepper photolithography at 248 nm wavelength, followed by an inductively coupled plasma reactive ion etching process. The waveguides were partially etched to 70 nm depth, and their width was 700 nm. The basic differential group delay unit $\tau $ was chosen as 0.37 ps (paths imbalance of 30 µm). This delay corresponds to a channel spacing of 0.96 nm (167 GHz). The device layout included an input port, 16 output ports, and test ports at the unused input ends of each of the 15 interferometer stages. The longest optical path between input and output was about 7 mm. Wafers were diced into 1.1 × 1.1 cm2 dies for subsequent processing and testing. The size of each 16-channel WDM device is 0.4 × 0.2 cm2.
To allow for phase delay trimming, patches of photosensitive As2Se3 chalcogenide glass were deposited on top of the silicon-photonic waveguides. Upper cladding layer regions were defined on the longer arm of each interferometer ( [37], see Fig. 1). The lengths of the chalcogenide glass cladding regions were 260 µm, 130 µm, 65 µm and 32.5 µm within stages A through D, and their thickness was 180 nm. The lengths of the upper cladding patches were proportional to the differential group delay within each interferometer stage. In this manner, the differential group delays remain integer multiples of a basic unit value and the distortion of transmission spectra is minimized.
The upper cladding layer regions were defined by photolithography, thermal evaporation, and liftoff. A bulk As2Se3 target was evaporated from a molybdenum boat at 1.2 × 10−5 mbar pressure. The deposition rate was 0.3 nm×s−1. The refractive index of the chalcogenide glass cladding layer at 1310 nm wavelength is 2.8 refractive index units (RIU) [38]. Numerical simulations suggest that the effective index of the single transverse-electric mode of the SOI waveguide at 1310 nm wavelength is 2.99 RIU and 2.87 RIU, with and without the chalcogenide glass upper cladding, respectively. Figure 3 shows a top-view microscope image of a fabricated device, with patches of photosensitive upper cladding.
Fig. 3. Top-view optical microscope image of a 16-channel, wavelength division multiplexing device in silicon-on-insulator. The scale bar corresponds to 400 µm. Patches of an upper cladding layer of photosensitive As2Se3 chalcogenide glass may be observed.
The setup for characterization of WDM devices is illustrated in Fig. 4(a). Light from a broadband source in the O-band wavelengths range was coupled into the device under test using a standard single-mode fiber. The input fiber was positioned on top of a vertical grating coupler at the edge of either the input port waveguide or one of the test ports waveguides (see Fig. 1), using a six-axis micro-mechanical precision stage. The input fiber was moved among the input and test ports during the testing protocol, as necessary. A second standard single-mode fiber was placed above a vertical grating coupler at the edge of one of the output port waveguides of the device, using a second six-axis stage. The output fiber was also moved among the different output ports during testing. The facets of input and output fibers were cleaved at 40° angles, so that guided light could be folded to/from the grating couplers with the fibers held in parallel to the device surface. Collected light was measured using an optical spectrum analyzer. The transmission of optical power between input and output ports was evaluated based on the output optical spectrum. A reference trace was acquired by connecting the broadband source directly to the spectrum analyzer. This trace was used for calibration of transmission losses.
Fig. 4. (a): Schematic illustration of the experimental setup used in the characterization of wavelength division multiplexing devices and for the trimming of differential phase delays. (b): Image of the device under test within the setup. The objective lens is used to focus the beam of an ultrafast Ti:Sapphire laser on the device surface. The beam can locally remove the upper cladding layer of photosensitive chalcogenide glass from the illuminated waveguide segments, thereby changing the effective index of the waveguide. (c): Top view optical microscope images of a chalcogenide glass upper cladding patch on top of a silicon waveguide core, within a device under test. The scale bars correspond to 10 µm. The left image shows the patch before the trimming procedure, whereas the right image shows the same patch after partial mass transfer during the trimming procedure. The location of the focused Ti:Sapphire laser beam with respect to the edge of the upper cladding region is illustrated by a white circle in the right image. The direction of a line scan of the beam position, perpendicular to the waveguide, is noted by a white arrow. Repeating line scans were performed to remove the upper cladding from waveguide segments of short incremental lengths.
Post-fabrication trimming of phase delays within interferometer stages was carried out by localized illumination of the upper As2Se3 cladding regions by the focused beam of a Ti:Sapphire ultrafast laser [37]. The laser emitted 200 fs-long pulses with a repetition rate of 80 MHz. The central wavelength of the laser was adjusted to 810 nm, and its average intensity was attenuated to 3 MW×cm−2. The laser beam was focused on the surface of the device using an objective lens. The spot size diameter was 2 µm. The image of a device within the phase trimming setup is shown in Fig. 4(b). The sample under test, together with the input and output fibers and their positioning stages, were scanned through the focused laser beam. Repeating line scans were performed perpendicular to the waveguide (see Fig. 4(c)), with scanning velocity of 2.9 µm×s−1.
Illumination resulted in the local mass transfer of the upper cladding layer [32,33]. The exposure of the underlying silicon core to air modified the effective index of the irradiated waveguide segment. The beam was scanned in partial overlap with the edge of the upper cladding patch (Fig. 4(c)), to allow for the exposure of short waveguide segments. The upper cladding layer was removed from 0.7 µm-long waveguide increments, corresponding to phase delay variation steps of about $2\pi /15$ rad. Following each photo-removal step, the transfer function of optical power between the pair of input and output ports was measured using the broadband source and the optical spectrum analyzer, and the results were compared with a pre-calculated theoretical model. Additional waveguide segments were irradiated as necessary, until the differential phase delay of the interferometer stage under study reached its target value. Then, the input and output fibers were moved to address another interferometer stage using the same protocol, until the response of the entire device was successfully adjusted.
Figure 5(a) presents measured normalized transfer functions of optical power between the common input port and one output port. The different traces show the initial response before the trimming protocol (blue), and additional measurements taken following the sequential trimming of phase delays within interferometer stages D8 (orange), C4 (yellow), B2 (magenta), and A (green). The differential phase delays in the as-fabricated device were arbitrary, leading to large deviations between the measured transfer function (blue) and the design (dashed). At the conclusion of the process (green), the transfer function matched the design of a 16-channel WDM device (dashed). In the specific device under test, the B2 and A interferometer stages had as-fabricated differential phases close to the target designs. Thus, trimming resulted in little residual improvement on the transfer function shown. However, the initial as-fabricated differential phases are arbitrary and their agreement with pre-selected values cannot be relied upon.
Fig. 5. (a): Measured normalized transfer functions of optical power from the common input port of a 16-channel WDM device and one output port. Traces were acquired before the trimming of phase delays (blue), and again following the sequential adjustments of phase delays within interferometer stages D8 (orange), C4 (yellow), B2 (magenta), and A (green). At the conclusion of the process, the transfer function matched the design of a channel selection filter (dashed line). (b): Solid lines - Measured normalized transfer functions of optical power from the common input port to the 16 output ports, following the trimming of differential phase delays in the entire WDM device. Dashed lines – corresponding modelled transfer functions. Channel numbers are noted above the panel.
Figure 5(b) (solid lines) shows the measured normalized transfer functions ${|{{H_m}(\lambda )} |^2}$ of optical power to all 16 output ports following the trimming of phase delays over the entire device. Here $\lambda $ stands for wavelength. The dashed lines show the corresponding modelled transfer functions. Results are shown within a single spectral period of the transfer functions. The wavelength demultiplexing of input channels to the different physical output ports is successfully realized. The spacing between adjacent channels is 0.96 nm, and the free spectral range of the responses is 15.44 nm. The full width $\Delta \lambda $ at −1 dB transmission of each of the 16 channels is 0.5 nm (87 GHz). The losses of optical power between input and outputs at the wavelengths of maximum transmission of each port were 23 ± 1 dB. The losses are dominated by the efficiency of coupling between fiber and waveguide at the vertical grating couplers, on the order of 10 dB per facet.
To quantify the channel de-multiplexing performance, we define the following set of coefficients:
In Eq. (2) ${\lambda _k}$ is the peak transmission wavelength of output port k. For $m = k$, ${S_{mm}}$ denotes the integrated power of a communication signal that is successfully routed to its designed output port. For $m \ne k$, ${S_{mk}}$ represents a contribution of inter-channel crosstalk. It is assumed that no signals are transmitted at the guard bands outside the wavelength regions $\left[ {\begin{array}{{cc}} {{\lambda_k} - \mathrm{\Delta }\lambda /2}&{{\lambda_k} + \mathrm{\Delta }\lambda /2} \end{array}} \right]$. The inter-channel crosstalk at output channel m may be estimated as [39]:
The experimental crosstalk ratios of the 16 demultiplexed channels of Fig. 5(b) are 12.1 ± 1.7 dB. The crosstalk performance that can be obtained with a single Mach-Zehnder interferometer in each filter stage and perfect adjustment of differential phases is 16 dB, as also recently shown in a 16-channel C-band device [40]. Better crosstalk performance requires multiple interferometers in each stage, as in [30]. Residual errors in differential phase delays at the conclusion of the trimming process therefore degrade the average crosstalk performance by 3.9 dB. For on-off keying transmission and a receiver that is restricted by additive thermal noise, the power penalty (dB) associated with inter-channel crosstalk is estimated as [39]:
Here $\bar{X}$ is the mean of inter-channel crosstalk values $\{{{X_m}} \}$ of the 16 output ports, and the Q factor represents the bit error ratio required at the receiver. For a bit error ratio of 10−9 ($Q$ = 6), the power penalty P associated with the measured inter-channel crosstalk is only 0.1 dB.
Residual errors in the splitting ratios of directional couplers may also contribute to crosstalk among the channels. The splitting ratios were estimated based on measurements of the transfer functions of the D-stages Mach-Zehnder interferometers. The ratios were within the range of 0.5 ± 0.01. The effect of this uncertainty on the performance of the entire WDM device is much smaller than that of residual errors in differential phase delays. Robust designs of directional couplers, which are less sensitive to fabrication tolerances, may be used as necessary [41]. Finally, the splitting ratios may also be trimmed through controlled illumination of a photosensitive upper cladding layer [36], using a similar protocol to the one used here in the adjustment of phase delays.
Table 2 compares the performance metrics of 12 recent reports of O-band WDM devices in silicon photonics, including this work. Most devices relied on cascaded Mach-Zehnder interferometer stages. The devices exhibit different tradeoffs between performance and simplicity of design and operation. The state-of-the-art has been recently extended by the work of Akiyama et al., who demonstrated the multiplexing of 32 channels with 50 GHz spacing and crosstalk levels below −38 dB [30]. This excellent performance was reached using 31 cascaded filter stages, each comprised of three interferometers. The tuning of the device required active thermal heating and feedback through 93 control elements. Compared with this work, the performance achieved in [30] is superior, at a cost of greater complexity: Following one-time trimming, the operation of the device reported herein is passive and control-free. Except for [20] and [30], all other previously reported devices supported 8 channels of fewer, while some such as [27] required active tuning. The 16-channel WDM device by Davis et al. exhibited better crosstalk than this work [20], but wider channels spacing and active thermal tuning.
Table 2. Comparison of silicon-photonic, O-band wavelength division multiplexing devices.
The device realized in this work was stored in the dark following the trimming process, to try and minimize unintended modifications to the transfer functions. The measurements of Fig. 5(b) were repeated one week following trimming, and again after two more weeks. The transfer functions were generally stable, with only small-scale drifting. The inter-channel crosstalk values changed to 10.0 ± 1.2 dB and 9.4 ± 1.2 dB after one week and three weeks since the trimming of phase delays, respectively. The corresponding power penalties remain modest: 1 dB and 1.4 dB. Five months following trimming, inter-channel crosstalk deteriorated further to 7.7 ± 1.7 dB. With that crosstalk level, bit error ratios of 10−9 cannot be supported directly, and forward error correction is necessary. Long-term drifting of the upper cladding layer is the subject of ongoing investigation.
4. Summary
In this work, we have designed and demonstrated a 16-channel WDM device for O-band operation in the standard SOI platform. The device consists of 15 Mach-Zehnder interferometers cascaded in a four-stage tree topology. The differential phase delays within each interferometer were trimmed post-fabrication through local illumination of a photosensitive upper cladding layer of As2Se3 chalcogenide glass. Trimming was carried out using a closed-loop protocol, with feedback of continuous measurements of power transfer functions between test port inputs and channel outputs. The channels spacing of 0.96 nm was arbitrarily chosen to demonstrate comparatively close spacing and was not meant to address a specific data communications standard.
The one-time post-fabrication trimming of phase delays allows for fully passive operation of the device, which is free of continuous feedback and active elements. The injection of currents or charges and the monitoring of feedback photodiodes is not necessary. Both the fabrication and the employment of the WDM device are therefore considerably simplified. The advantage of passive operation becomes more significant with increasing channel count, as the number of phase delay degrees of freedom becomes larger. Thermal tuning, in particular, is prone to crosstalk across the silicon device layer, which could make the adjustment of specific phase delays more challenging. The local photo-induced mass transfer of an upper cladding layer above a specific waveguide has no effect on other locations.
The power consumption of active thermal tuning is on the order of 10 mW per heating element [27]. The one-time trimming of phase delays may, in principle, save 150 mW of power during device operation. Note, however, that the transfer functions of the device are expected to drift with ambient temperature. The response of the device must be kept on a known wavelengths grid, through either a passive heat-sink or thermoelectric cooling. In the latter case, any power savings associated with one-time trimming of phase delays are negated. Nevertheless, the significant added value of simple, feedback-free operation of trimmed devices would remain.
The inter-channel crosstalk of the devices degraded by 2.7 dB over several months following one-time trimming. Measures for improving the long-term stability of the photosensitive cladding layer after trimming, and the compatibility of the layer with certain packaging protocols such as the ultraviolet light curing of adhesives, still require attention. The illumination of the upper cladding with a lower intensity laser beam can lead to a photo-darkening effect rather than mass transfer [32,33], resulting in finer tuning of differential phase delays [37] and lower residual crosstalk. Arsenic-free compositions of photosensitive glasses, such as GeSe and GeSe2 [42], are also being investigated.
In conclusion, the WDM devices presented in this work can be useful in silicon-photonic integrated circuits for large-volume data communications. The devices can be mass-produced at low cost, and their trimming protocol can be fully automated. Future work would involve the application of the devices within data centers.
Funding
Israel Innovation Authority (MAGNETON 68966).
Disclosures
The authors declare no conflict 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. G. T. Reed, Silicon Photonics: The State of the Art (Wiley, (2008).
2. L. Vivien and L. Pavesi Eds., Handbook of Silicon Photonics (Taylor & Francis, 2016).
3. D. A. B. Miller, “Device requirements for optical interconnects to silicon chips,” Proc. IEEE 97(7), 1166–1185 (2009). [CrossRef]
4. H. Lira, S. Manipatruni, and M. Lipson, “Broadband hitless silicon electrooptic switch for on-chip optical networks,” Opt. Express 17(25), 22271–22280 (2009). [CrossRef]
5. S. Ibrahim, N. K. Fontaine, S. S. Djordjevic, B. Guan, T. Su, S. Cheung, R. P. Scott, A. T. Pomerene, L. L. Seaford, C. M. Hill, S. Danziger, Z. Ding, K. Okamoto, and S. J. B. Yoo, “Demonstration of a fast-reconfigurable silicon CMOS optical lattice filter,” Opt. Express 19(14), 13245–13256 (2011). [CrossRef]
6. P. Dong, W. Qian, H. Liang, R. Shafiiha, D. Feng, G. Li, J. E. Cunningham, A. V. Krishnamoorthy, and M. Asghari, “Thermally tunable silicon racetrack resonators with ultralow tuning power,” Opt. Express 18(19), 20298–20304 (2010). [CrossRef]
7. J. Sun, E. Timurdogan, A. Yaacobi, E. S. Hosseini, and M. R. Watts, “Large-scale nanophotonic phased array,” Nature 493(7431), 195–199 (2013). [CrossRef]
8. F. Horst, W. M. J. Green, S. Assefa, S. M. Shank, Y. A. Vlasov, and B. J. Offrein, “Cascaded Mach-Zehnder wavelength filters in silicon photonics for low loss and flat pass-band WDM (de-)multiplexing,” Opt. Express 21(10), 11652–11658 (2013). [CrossRef]
9. S.-H. Jeong, D. Shimura, T. Simoyama, M. Seki, N. Yokoyama, M. Ohtsuka, K. Koshino, T. Horikawa, Y. Tanaka, and K. Morito, “Low-loss, flat-topped and spectrally uniform silicon nanowire-based 5th-order CROW fabricated by ArF-immersion lithography process on a 300-mm SOI wafer,” Opt. Express 21(25), 30163–30174 (2013). [CrossRef]
10. T. Fukazawa, F. Ohno, and T. Baba, “Very compact arrayed-waveguide grating demultiplexer using Si photonic wire waveguides,” Jpn. J. Appl. Phys. 43(5B), L673–L675 (2004). [CrossRef]
11. P. Cheben, J. H. Schmid, A. Delâge, A. Densmore, S. Janz, B. Lamontagne, J. Lapointe, E. Post, P. Waldron, and D.-X. Xu, “A high-resolution silicon-on-insulator arrayed waveguide grating microspectrometer with submicrometer aperture waveguides,” Opt. Express 15(5), 2299–2306 (2007). [CrossRef]
12. S. Pathak, M. Vanslembrouck, P. Dumon, D. Van Thourhout, and W. Bogaerts, “Optimized silicon AWG with flattened spectral response using an MMI aperture,” J. Lightwave Technol. 31(1), 87–93 (2013). [CrossRef]
13. S. Cheung, T. Su, K. Okamoto, and S. J. B. Yoo, “Ultra-compact silicon photonic 512 × 512 25 GHz arrayed waveguide grating router,” IEEE J. Sel. Top. Quantum Electron. 20(4), 310–316 (2014). [CrossRef]
14. S. Pathak, P. Dumon, D. Van Thourhout, and W. Bogaerts, “Comparison of AWGs and Echelle gratings for wavelength division multiplexing on silicon-on-insulator,” IEEE Photonics J. 6(5), 1–9 (2014). [CrossRef]
15. J. Brouckaert, W. Bogaerts, P. Dumon, D. Van Thourhout, and R. Baets, “Planar concave grating demultiplexer fabricated on a nanophotonic silicon-on-insulator platform,” J. Lightwave Technol. 25(5), 1269–1275 (2007). [CrossRef]
16. F. Horst, W. M. J. Green, B. Offrein, and Y. Vlasov, “Silicon-on-insulator echelle grating WDM demultiplexers with two stigmatic points,” IEEE Photonics Technol. Lett. 21(23), 1743–1745 (2009). [CrossRef]
17. X. Pommarede, K. Hassan, P. Billondeau, V. Hugues, P. Grosse, B. Charbonnier, and N. S. G.-H. Duan, “16 times 100 GHz Echelle grating-based wavelength multiplexer on silicon-on-insulator platform,” IEEE Photonics Technol. Lett. 29(6), 493–495 (2017). [CrossRef]
18. A. V. Krishnamoorthy, X. Zheng, D. Feng, J. Lexau, J. F. Buckwalter, H. D. Thacker, F. Liu, Y. Luo, E. Chang, P. Amberg, I. Shubin, S. S. Djordjevic, J. H. Lee, S. Lin, H. Liang, A. Abed, R. Shafiiha, K. Raj, R. Ho, M. Asghari, and J. E. Cunningham, “A low-power, high-speed, 9-channel germanium-silicon electro-absorption modulator array integrated with digital CMOS driver and wavelength multiplexer,” Opt. Express 22(10), 12289–12295 (2014). [CrossRef]
19. Z. Liu, J. Zhang, X. Li, L. Wang, J. Li, C. Xue, J. An, and B. Cheng, “25 × 50 Gbps wavelength division multiplexing silicon photonics receiver chip based on a silicon nanowire-arrayed waveguide grating,” Photonics Res. 7(6), 659–663 (2019). [CrossRef]
20. J. A. Davis, A. Li, N. Alshamrani, and Y. Fainman, “Silicon photonic chip for 16-channel wavelength division (de-)multiplexing in the O-band,” Opt. Express 28(16), 23620–23627 (2020). [CrossRef]
21. Z. Zhao, Z. Li, J. Niu, G. Zhang, H. Chen, X. Fy, and L. yang, “Eight-channel LAN WDM (De)multiplexer based on cascaded Mach–Zehnder interferometer on SOI for 400 GbE,” Photonics 9(4), 252 (2022). [CrossRef]
22. S.-H. Jeong, T. Horikawa, and T. Nakamura, “Phase behaviors for silicon-wire multistage delayed interferometric WDM filters across a whole 300-mm silicon-on-insulator wafer,” J. Opt. Soc. Am. B 37(6), 1847–1856 (2020). [CrossRef]
23. S.-H. Jeong, “Broadband 1×8 channel silicon-nanowire-waveguide WDM filter based on point-symmetric Mach-Zehnder interferometric optical couplers in the O-band spectral regime,” OSA Continuum 2(12), 3564–3575 (2019). [CrossRef]
24. H. Xu, D. Dai, and Y. Shi, “Low-crosstalk and fabrication-tolerant four-channel CWDM filter based on dispersion-engineered Mach-Zehnder interferometers,” Opt. Express 29(13), 20617–20631 (2021). [CrossRef]
25. L. Chang, Y. Gong, L. Liu, Z. Li, and Y. Yu, “Low-Loss Broadband Silicon-on-Insulator Demultiplexers in the O-Band,” IEEE Photonics Technol. Lett. 29(15), 1237–1240 (2017). [CrossRef]
26. T.-H. Yen and Y.-J. Hung, “Fabrication-tolerant CWDM (de)multiplexer based on cascaded Mach–Zehnder interferometers on silicon-on-insulator,” J. Lightwave Technol. 39(1), 146–153 (2021). [CrossRef]
27. N. Ning, X. Wang, H. Yu, Q. Huang, Y. Wang, and J. Yang, “Comparison of silicon lattice-filter-based O-band 1×8 (de)multiplexers with flat and Gaussian-like passbands,” IEEE Photonics J. 14(2), 1–5 (2022). [CrossRef]
28. H. Xu and Y. Shi, “Flat-top CWDM (de)multiplexer based on MZI with bent directional couplers,” IEEE Photonics Technol. Lett. 30(2), 169–172 (2018). [CrossRef]
29. S. Pitris, G. Dabos, C. Mitsolidou, T. Alexoudi, P. De Heyn, J. Van Campenhout, R. Broeke, G. T. Kanellos, and N. Pleros, “Silicon photonic 8 × 8 cyclic arrayed waveguide grating router for O-band on-chip communication,” Opt. Express 26(5), 6276–6284 (2018). [CrossRef]
30. T. Akiyama, M. Nishizawa, A. Sugama, Y. Nakasha, S. Tanaka, Y. Tanaka, and T. Hoshida, “First demonstration of crosstalk-free (<−38.5 dB) 32-ch DWDM demultiplexer on standard Si PIC platform,” paper Th4C-2, in Optical Fiber Communication Conference (OFC), Proc. Optica Publishing Group, (March 2022).
31. C. K. Madsen and J. H. Zhao, Optical Filter Design and Analysis – A Signal Processing Approach (Wiley, 1999).
32. A. E. Owen, A. P. Firth, and P. J. S. Ewen, “Photo-induced structural and physico-chemical changes in amorphous chalcogenide semiconductors,” Philos. Mag. Part B 52(3), 347–362 (1985). [CrossRef]
33. A. Zoubir, M. Richardson, C. Rivero, A. Schulte, C. Lopez, K. Richardson, N. Hô, and R. Vallée, “Direct femtosecond laser writing of waveguides inAs2 S3 thin films,” Opt. Lett. 29(7), 748–750 (2004). [CrossRef]
34. A. Canciamilla, F. Morichetti, S. Grillanda, P. Velha, M. Sorel, V. Singh, A. Agarwal, L. C. Kimerling, and A. Melloni, “Photo-induced trimming of chalcogenide-assisted silicon waveguides,” Opt. Express 20(14), 15807–15817 (2012). [CrossRef]
35. R. Califa, Y. Kaganovskii, D. Munk, H. Genish, I. Bakish, M. Rosenbluh, and A. Zadok, “Large photo-induced index variations in chalcogenide-on-silicon waveguides,” Opt. Lett. 39(20), 5905–5908 (2014). [CrossRef]
36. R. Califa, D. Munk, H. Genish, Y. Kaganovskii, I. Bakish, M. Rosenbluh, and A. Zadok, “Large one-time photo-induced tuning of directional couplers in chalcogenide-on-silicon platform,” Opt. Express 23(22), 28234–28243 (2015). [CrossRef]
37. D. Munk, M. Katzman, Y. Kaganovskii, N. Inbar, A. Misra, M. Hen, M. Priel, M. Feldberg, M. Tkachev, A. Bergman, M. Vofsi, M. Rosenbluh, T. Schneider, and A. Zadok, “Eight-Channel silicon-photonic wavelength division multiplexer with 17 GHz spacing,” IEEE J. Sel. Top. Quantum Electron. 25(5), 1–10 (2019). [CrossRef]
38. R. Todorov, J. Tasseva, T. Babeva, and K. Petkov, “Multilayer As2Se3/GeS2 quarter wave structures for photonic applications,” J. Phys. D: Appl. Phys. 43(50), 505103 (2010). [CrossRef]
39. P. Agrawal, Fiber-Optic Communication Systems, 3rd ed. (Wiley Interscience, 2002).
40. L. Han, B. P. P. Kuo, A. Pejkic, N. Alic, and S. Radic, “50 GHz silicon cascaded Mach-Zehnder wavelength filter and automatic phase error correction,” paper W3B-3, in Optical Fiber Communication Conference (OFC), Proc. Optica Publishing Group, (March 2019).
41. M. Katzman, Y. Piasetzky, E. Rubin, B. Barenboim, M. Priel, M. Erew, A. Zadok, and A. Suchowski, “Robust directional couplers for state manipulation in silicon photonic-integrated circuits,” J. Lightwave Technol. 40, (in press, 2022).
42. R. Schardt, “Photodarkening of germanium-selenium glasses induced by below-bandgap light,” Ph.D. Dissertation, Univ. of Florida (2000).
