Angled bent multimode interferometer for coarse wavelength division multiplexing
The ABMMI design addresses phase errors in CWDM filters by optimizing parameters like Rmmi and θIO, enhancing self-image quality and reducing crosstalk, resulting in efficient and compact optical filters.
Patent Information
- Application Number
- PCT/CN2025/082495
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2024-03-14
- Filing Date
- 2025-03-14
- Publication Date
- 2025-09-18
AI Technical Summary
Existing AMMI-based CWDM filters suffer from phase error issues leading to high insertion loss and crosstalk, complicating fabrication and increasing device footprint, especially when used with other interleavers like MZIs.
An angled bent multimode interferometer (ABMMI) design with optimized parameters such as inner radius (Rmmi), mode order range, and angles (θIO) to form high-quality self-images, reducing phase error and crosstalk without increasing device size.
Achieves low insertion loss (0.25-0.71 dB) and crosstalk (-19.0 to -27.8 dB) with improved self-image fidelity, maintaining a compact footprint and tolerance to laser wavelength drift.
Smart Images

Figure CN2025082495_18092025_PF_FP_ABST
Abstract
Description
ANGLED BENT MULTIMODE INTERFEROMETER FOR COARSE WAVELENGTH DIVISION MULTIPLEXING
[0001] CROSS-REFERENCE TO RELATED APPLICATION
[0002] The present application claims the benefit of U.S. Provisional Application Serial No. 63 / 565,360, filed March 14, 2024, the disclosure of which is incorporated herein by reference in its entirety.BACKGROUND OF THE INVENTION
[0003] I. INTRODUCTION
[0004] Optical interconnect is a crucial technology meeting ever-increasing demand in intra-data center transmission capacity driven by bandwidth-intensive applications, including high performance computing (HPC) and large-scale artificial intelligence (AI) models [1, 2] . Adopting the CWDM technique not only amplifies link capacity by allowing multiple data carriers with 20 nm wavelength spacing transmitted over the same fiber, but also benefit from cost-effectiveness in terms of relaxing the precise laser wavelength control that is typically required by dense WDM (DWDM) and thus achieving higher laser production yield [3] .
[0005] On-chip CWDM filters are vital devices for implementing the CWDM technique and have been demonstrated by making use of Bragg gratings [4, 7] , arrayed waveguide gratings (AWGs) [5] , cascaded Mach-Zehnder interferometers (CMZIs) [8-12] and Angled Multimode Interferometers (AMMIs) [13-19] . Among these, AMMIs can be utilized to be CWDM filters by harnessing the dispersive self-imaging effect in multimode waveguides, which is because AMMIs have the advantages of easy fabrication with a single etching step and being more fabrication tolerant than AWG-based [5, 6] and normal CMZI-based [8, 10] design. Previously demonstrated AMMI for four wavelength channels with 20 nm spacing
[0013] is able to achieve moderate crosstalk (< -15dB) at channel central wavelengths for four-channel 20 nm spacing design. However, the AMMI-based CWDM filters inherently suffer from the phase error problem in MMI [13, 20] , which means the real effective indices of the excited higher order modes deviate more significantly from ideal perfect-image-forming effective indices, leading to deteriorated self-image fidelity. This gives rise to a relatively high insertion loss (2 dB) and strong transmission spectral side lobes of each channel, resulting in limited low-crosstalk bandwidth
[0013] . Moreover, the spectral side lobes of the AMMI will cause more significant side lobe problems of the whole filter system when the AMMI is utilized together with other interleavers such as MZIs
[0014] . Therefore, to further improve the insertion loss and crosstalk performance of the AMMI-based CWDM filter, it is necessary to solve the phase error problem.
[0006] To reduce the phase error in the MMI, there have been pre-existing solutions based on reducing the effective index contrast between the MMI core and clad region. [15, 21-23] effectively reduce the phase error for higher order modes and improve the self-image quality by making use of cladding with shallow etching depth. However, the method complicates the fabrication process as two-step etching is required, and it may need the custom etching depth, which is different from the MPW service provided by the foundry for optimized design. Also, the performance can be sensitive to the etching depth
[0021] .
[0022] utilizes subwavelength grating cladding to lower the core-cladding index contrast and improve the self-image quality, which only needs one step of etching. Still, the subwavelength structure could be challenging to fabricate, especially when the patterned subwavelength cladding needs to maintain stability over a long AMMI length (>1 mm) for wavelength demultiplexing. Besides, the mode orders excited by angled input for CWDM usage are typically higher than those in the above work, where the mode orders are excited by normal input. In addition, it should be noted that by using an AMMI filter with a wider MMI section and input / output waveguides [14, 15] , the excited modes could be concentrated more on lower orders, resulting in decreased phase error.
[0007] However, this leads to the emergence of a significantly elongated MMI (>3 mm) , given that the self-image length increases quadratically with respect to MMI width, resulting in both higher insertion loss and larger footprints. Additionally, the adiabatic tapers connecting input / output waveguides and single mode waveguides for on-chip routing would necessitate extension. Therefore, there is a pressing need for additional solutions to mitigate the phase error issue in MMI-based filter design without introducing complexities to the fabrication process or further enlarging the device footprint.
[0008] There continues to be a need in the art for improved designs and techniques for a method and systems for coarse wavelength division multiplexing.
[0009] BRIEF SUMMARY OF THE INVENTION
[0010] According to an embodiment of the subject invention, a Coarse Wavelength Division Multiplexing (CWDM) filter system based on an angled bent multimode interferometer (ABMMI) is provided, comprising one input port; a plurality of output ports; and a bent multimode waveguide connecting the input port and the plurality of output ports, having an inner radius of Rmmi; wherein light incident to the input port is demultiplexed into a plurality of wavelength channels. Each of the input port and plurality of output port comprises a first waveguide having a width WIO and a single mode waveguide having a width Wsm connected to the first waveguide, wherein the width WIO is greater than the width Wsm. The plurality of output ports have four output ports. Moreover, each of the plurality of output ports corresponds to a channel and the demultiplexed light is output from the plurality of channels, respectively. The incident light demultiplexed into the plurality of wavelength channels has center wavelengths of 1591nm, 1571nm, 1551nm and 1531nm, respectively. The inner radius Rmmi is optimized to form a high-quality self-image at the output ports. An excited mode order range of the CWDM filter system is optimized to form a high-quality self-image at the output ports. Further, an angle between a middle portion of respective input port and the output portions and tangent of an inner arc of the bent multimode waveguide is defined as θIO. θIO and WIO are optimized for sufficient distances between adjacent output ports for channel spacing such that a minimal crosstalk between output wavelength channels is obtained. Insertion Loss (IL) in a range between 0.25 dB and 0.71 dB and Crosstalk (XT) in a range between -19.0 dB and -27.8 dB at channel central wavelengths are obtained.
[0011] In certain embodiment of the subject invention, a method for optimizing parameters of a Coarse Wavelength Division Multiplexing (CWDM) filter system based on angled bent multimode interferometer (ABMMI) comprising one input port; a plurality of output ports; and a bent multimode waveguide connecting the input port and the plurality of output ports, having an inner radius of Rmmi; wherein light incident to the input port is demultiplexed into a plurality of wavelength channels is provided. The method comprises optimizing the inner radius Rmmi to form a high-quality self-image at the output ports. The method may further comprise optimizing an excited mode order range of the CWDM filter system to form a high-quality self-image at the output port. The input port and the output ports each has a width WIOand an angle between a middle portion of respective input / output and tangent of an inner arc of the bent multimode waveguide is defined as θIO. In a preferred embodiment, the input port and the output ports each has a same width to minimize the optical loss. Nevertheless, the input port and the output ports may have different widths. The method may further comprise optimizing θIO and WIO for sufficient distances between adjacent output ports for channel spacing such that a minimal crosstalk between output wavelength channels is obtained. Insertion Loss (IL) in a range between 0.25 dB and 0.71 dB and Crosstalk (XT) in a range between -19.0 dB and -27.8 dB at channel central wavelengths are obtained.
[0012] In another embodiment of the subject invention, a non-transitory computer readable medium having stored therein program instructions executable by a computing system to cause the computing system to perform a method for optimizing parameters of a Coarse Wavelength Division Multiplexing (CWDM) filter system based on angled bent multimode interferometer (ABMMI) comprising one input port; a plurality of output ports; and a bent multimode waveguide connecting the input port and the plurality of output ports, having an inner radius of Rmmi; wherein light incident to the input port is demultiplexed into a plurality of wavelength channels, the method comprising optimizing the inner radius Rmmi to form a high-quality self-image at the output ports. The method may further comprise optimizing an excited mode order range of the CWDM filter system to form a high-quality self-image at the output ports. The input port and the output ports each has a width WIO and an angle between a middle portion of respective input port and output ports and tangent of an inner arc of the bent multimode waveguide is defined as θIO. The method may further comprise optimizing θIO and WIO for sufficient distances between adjacent output ports for channel spacing such that a minimal crosstalk between output wavelength channels is obtained. Insertion Loss (IL) in a range between 0.25 dB and 0.71 dB and Crosstalk (XT) in a range between -19.0 dB and -27.8 dB at channel central wavelengths are obtained.BRIEF DESCRIPTION OF THE DRAWINGS
[0013] FIG. 1A is a schematic illustration of the ABMMI-based CWDM filter according to an embodiment of the subject invention, wherein the inset shows the cross-section of the device of the embodiments of the subject invention, and FIG. 1B is a schematic illustration of the conventional AMMI-based CWDM filter according to prior art.
[0014] FIG. 2 is diagram of the p value of the parabolic Equation (3) fitted by Equation (9) versus Rmmi for different mode order ranges used for fitting, wherein markers in the shape of diamond in Figure 2 show the possible working points, and red marker shows the working point used for the designed ABMMI, according to an embodiment of the subject invention.
[0015] FIGS. 3A, 3C, and 3E show results of analysis of effective indices relation and phase error in the MMWG region for bent MMI and FIGS. 3B, 3D, and 3F show results of analysis of effective indices relation and phase error in the MMWG region for conventional straight MMI, wherein FIGS. 3A and 3B show the simulated effective indices and corresponding quadratic fitting, FIGS. 3C and 3D show the phase error for two types of MMWG defined by (10) , FIGS. 3E and 3F are zoom-in view of FIGS. 3C and 3D, respectively; wherein the region between vertical red dashed lines indicates the order region of excited modes that includes >97%input power, according to an embodiment of the subject invention.
[0016] FIG. 4A shows power ratio of modes in the bent MMI region coupled by fundamental mode in the input waveguide at different input waveguide angle θIO, wherein FIG. 4B shows power ratio diagram sliced from FIG. 4A at selected θIO=13°, and wherein corresponding cumulative power ratio diagram is also shown, according to an embodiment of the subject invention.
[0017] FIG. 5A shows simulated maximum transmission with respect to output port location maxL T verses Rmmi and θIO at 1561nm, wherein green star marks the design point, FIG. 5B shows illustration of the simulation method of the transmission and spectra of the whole device, FIG. 5C shows simulated transmission of the input / output tapers verses taper length LtIO at 1561nm, wherein vertical dashed line indicates the chosen taper length, according to an embodiment of the subject invention.
[0018] FIG. 6A shows simulated transmission spectra for the ABMMI according to an embodiment of the subject invention and FIG. 6B shows simulated transmission spectra for the conventional AMMI according to prior art, wherein horizontal dashed lines indicate the -3 dB and -20 dB transmission lines and vertical dashed lines indicate the four channel central wavelengths.
[0019] FIG. 7 shows results of fabrication tolerance analysis in terms of MMI waveguide width of the ABMMI, according to an embodiment of the subject invention.
[0020] FIG. 8 shows a microscope image of the fabricated device, according to an embodiment of the subject invention.
[0021] FIG. 9 shows measured transmission spectra of the fabricated ABMMI based CWDM filter, according to an embodiment of the subject invention.
[0022] TABLE I
[0023] a. Device / platform / working wavelength band.
[0024] b. Device size is written into the format Wmmi× (Lall+WmmiLall / (2Rmmi) ) =12×935.0 μm2 for easy comparison, where Lall=LID+LIOarc.
[0025] c. ILs at channels central wavelengths.
[0026] d. XTs at channels central wavelengths.
[0027] e. Sensitivities of central wavelength Δλ / ΔWmmi.
[0028] f. EFR, extra fabrication requirements.
[0029] DETAILED DISCLOSURE OF THE INVENTION
[0030] The embodiments of subject invention pertain to a Coarse Wavelength Division Multiplexing (CWDM) filter system and methods based on an angled bent multimode interferometer (ABMMI) .
[0031] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used herein, the term “and / or” includes any and all combinations of one or more of the associated listed items. As used herein, the singular forms “a, ” “an, ” and “the” are intended to include the plural forms as well as the singular forms, unless the context clearly indicates otherwise. It will be further understood that the terms “comprises” and / or “comprising, ” when used in this specification, specify the presence of stated features, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, steps, operations, elements, components, and / or groups thereof.
[0032] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one having ordinary skill in the art to which this invention pertains. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and the present disclosure and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
[0033] When the term “about” is used herein, in conjunction with a numerical value, it is understood that the value can be in a range of 90%of the value to 110%of the value, i.e. the value can be + / -10%of the stated value. For example, “about 1 kg” means from 0.90 kg to 1.1 kg.
[0034] In describing the invention, it will be understood that a number of techniques and steps are disclosed. Each of these has individual benefits and each can also be used in conjunction with one or more, or in some cases all, of the other disclosed techniques. Accordingly, for the sake of clarity, this description will refrain from repeating every possible combination of the individual steps in an unnecessary fashion. Nevertheless, the specification and claims should be read with the understanding that such combinations are entirely within the scope of the invention and the claims.
[0035] The Coarse Wavelength Division Multiplexing (CWDM) filters have extensive potential in data center communications, with the Angled Multimode Interferometer (AMMI) based filter offering a solution that demands only a single-step lithography and boasts relatively high fabrication tolerance. Nevertheless, enhancing the Insertion Loss (IL) and Crosstalk (XT) performance of an AMMI filter remains a formidable challenge without significantly increasing the device footprint.
[0036] A C-band four-channel CWDM filter based on angled bent MMI (ABMMI) is provided, achieving IL in a range between 0.25 dB and 0.71 dB and XT in a range between -19.0 and -27.8 dB at channel central wavelengths. Bandwidths for XT lower than -15 dB were characterized to be in range between 10.9 nm and 12.6 nm for four channels. The performance of the ABMMI is improved by alleviating the phase error problem by introducing a new degree of freedom, the radius of the MMI waveguide, and formulating a novel relaxed phase error optimization target. Thanks to these design methods, the device can be implemented with a single full etching process and is relatively compact in footprint compared to the existing AMMI-based design. Furthermore, up to date, this is the first time the self-imaging effect in a silicon bent multimode waveguide has been quantitatively explored and harnessed.
[0037] Device Design and Analysis
[0038] A. Self-image-forming Mechanism of Conventional AMMIs
[0039] For the conventional AMMI-based CWDM devices as shown in Fig. 1B, the wavelength demultiplexing is achieved by the difference of self-image location in the middle straight multimode waveguide region. Specifically, the location of the first inverted self-image is given by Equation (1) as shown in [13, 20]
[0040] where Lπ is the beating length between TE0 and TE1 modes, ni is the effective index of TEi mode (i=0, 1, 2…) in the middle straight multimode waveguide, k0=2π / λ0 is free-space propagation constant, λ0 is free-space wavelength, nr is the effective index of the waveguide core region and Wmmi is the width of the multimode waveguide. It can be seen from (1) that the self-image location varies with wavelength, by which the wavelength demultiplexing of the input light can be achieved. It should be noted that a perfect self-image can be theoretically formed at Ls if the effective indices of all the excited modes in the multimode waveguide follows the parabolic law as shown in Equation (2) :
[0041] which is an approximation of the real effective indices in the multimode waveguide
[0023] . However, the above approximation leads to the phase error problem in MMI self-imaging design, which is stated as follows. As the mode order i gets larger, the real effective indices ni deviate more severely from the ideal effective indices given by the above parabolic law, leading to more significant difference between real and ideal phases of the excited higher order modes. The high phase error will lower the quality of the formed self-image at the output waveguides, which gives rise to high loss due to the low mode overlap between the self-image and input / output TE0 mode profile, as well as high crosstalk and strong spectral side lobes due to power leakage to adjacent output ports.
[0042] Owing to the limitations of previous phase error controlling methods mentioned before, the phase error problem is alleviated by, first, introducing a generalized effective indices relation for forming the self-image and second, utilizing a bent multimode waveguide as the MMI region to satisfy this generalized effective indices relation.
[0043] B. Self-image-forming Mechanism of the ABMMI
[0044] The generalized effective indices relation is formulated as shown in Equation (3) :
[0045] where the coefficient Ls , p, and nC jointly determines the above parabolic low. where is the integer set, Ls can be any constant which will be shown to be the first self-image location later, and nC can be any constant with the unit same as an effective index. The relative phase of the excited ith mode at location Ls is denoted as shown in Equation (4) :
[0046] If the excited modes in the designed multimode waveguide follow (3) , then is deduced to be and as shown in Equation (4) :
[0047] Which means at location Ls, the first self-image is formed. When p is even, the first self-image is the mirrored image of input, whereas when p is odd, the first self-image is the same as input. It can be seen that (3) describes a larger set of image-forming parabolic relations compared with (2) , which making it easier to engineer the real ni in the multimode waveguide to approach in (3) , for the following reasons. First, coefficient p could be any integers instead of being restricted to p=2 as required in (2) . Second, the constant coefficient nC in (3) can be any value, whereas that in (2) is limited to be the effective index of the fundamental mode. In fact, (2) has been shown to be effective and can result in limited phase error in MMI design for power splitting
[0021] and 90° hybrid
[0022] where normal input is utilized and most of the input power is coupled into lower order modes in MMI region. However, in this case with angled input, the mode orders excited tend to be higher, such that good fitness between ni and is only required for higher i, instead of i being lower and starting from i=0.
[0048] C. Parameter Selection of Rmmi and Excited Mode Order Range
[0049] A new degree of freedom, the radius of the MMI section Rmmi, is provided to form a high-quality self-image and improve the loss and crosstalk performance. Fig. 1A is the schematic illustration of the ABMMI of the embodiments of the subject invention. The device comprises one input port, a plurality of, for example, four, output ports and a bent multimode waveguide with inner radius of Rmmi in the middle. The input and output waveguides with widths of WIO are connected to narrow single mode waveguides with widths of Wsm for on-chip routing by adiabatic tapers. The angles between the middle axes of the input / output tapers and the tangent of the multimode waveguide inner arc (indicated by dashed arc) are θIO. The incident light is demultiplexed into four wavelength channels, with center wavelengths 1591nm, 1571nm, 1551nm and 1531nm, and then output from channel 1, 2, 3 and 4, respectively. Considering that the designed device is working for four-channel coarse wavelength division multiplexing communications with 20nm channel spacing and enough distance should be ensured between adjacent output ports, the width of the bent MMI region is firstly selected to be 12μm. The optimization of the radius of the MMI section, Rmmi, is shown as follows. It can be seen from (3) that, forming a high-quality self-image can be achieved by engineering and fitting the ni of the excited modes closely to a relation, which is not only parabolic, but also the p coefficient of the parabolic relation is an integer. Also, it should be noted that, when p is chosen to be an even integer, the first self-image is formed by even-order modes and mirrored odd-order modes
[0020] . Since the mode profiles of the odd-order modes in bent waveguides do not show anti-symmetry as in straight waveguides, the first self-image is deformed from the input. Thus, p coefficient should be limited to an odd integer in the bent MMI case. For real effective indices ni in a bent multimode waveguide, the p coefficient can be deduced by the quadratic regression as shown in Equations (6) and (7) : minz||y-Mz||2 (6)
[0050] and
[0051] where i1…iN stands for the mode orders for the N excited modes in the bend MMI section, y denotes the vector composed by the real effective indices of these N modes, which can be obtained by finite difference element (FDE) , and z is the coefficient of the parabolic relation to be fitted. Compared with (3) , it can be seen as shown in Equation (8) :
[0052] By solving Equation (6) and using Equation (8) , parameters p and Ls can be expressed as Equation (9) :
[0053] Fig. 2 shows the fitted p value verses Rmmi for various initial excited mode order i1 at 1561nm wavelength. N is chosen to be 10 to ensure that >97%of the input power are coupled into TEi1 ~ TEiN modes, as the number of modes where most of input power is coupled into is close to 10, shown by later calculation in Fig. 4A. Points where the fitted p values are odd integer are highlighted in Fig. 2. When the radius of the MMI and the excited modes orders lies in the corresponding range of [i1, iN] , a high-quality self-image may be formed at the output port. The point (marked as orange) with Rmmi= 394μm and p = -1 are taken as the design point of the ABMMI structure of the embodiments of the subject invention, and it can be seen that the mode order range [i1, iN] with good fitness between ni and is [6, 15] .
[0054] To illustrate the improved image quality, Figs. 3A-3F compare real-ideal effective indices fitness for straight and bend multimode waveguide with Rmmi=394μm, respectively. Figs. 3C and 3E also show the phase error calculated by Equation (10) :
[0055] where the self-image location Ls is calculated by (1) and (8) for straight and designed bent MMI section, respectively. For the bent MMI, it can be observed that for i∈ [6, 15] , good fitness between ni and can be achieved, and the phase error can be maintained to be smaller than 10°, indicating that a high-quality self-image can be formed if input power is coupled into range TE6-TE15. Whereas for straight MMI, the phase error is larger than 10° for i>5 and increases rapidly with i, indicating that unless the mode order excited is controlled smaller than or equal to 5, the self-image would be severely blurred, especially for angled input that will obviously excite more higher order modes.
[0056] D. Parameter Selection of θIO and WIO
[0057] θIO and WIO are firstly decided to ensure enough distance between adjacent output ports for Δλ=20nm channel spacing, which leads to low crosstalk between output wavelength channels. The arc length occupied by one output port can be approximated by LIOarc≈WIO / sin (θIO) since it is far shorter than Rmmi. Thus, the above requirement can be formulated as shown in Equation (11) :
[0058] where the self-image length Ls is given by (9) . At λ=1561nm, the right-hand side of (11) is calculated to be 15.75 μm. WIO is chosen to be 3.2μm, thus, the feasible range for θIO should be θIO>11.72°. θIO is further optimized to make sure the excited modes belong to low-phase-error mode order range. The power ratio of the modes coupled from input port verses mode orders for various θIO is calculated by the Finite Difference Time Domain (FDTD) method, as shown in Fig. 4A. θIO is finally chosen to be 13° to make sure greater than 97%of the input power is coupled into mode order i∈ [6, 15] , which is shown in Fig. 4B.
[0059] E. Numerical validation of the Selected Parameters
[0060] To validate the parameters chosen above, Fig. 5A shows the simulated whole device maximum transmission with respect to output port location, maxL T , verses Rmmi and θIOat a wavelength of 1561 nm. The simulation scheme for the whole device is shown in Fig. 5B. The whole simulation is conducted by jointly using the FDTD and Eigenmode Expansion (EME) methods. The field pattern at interface I excited by input fundamental mode is calculated by the FDTD method. Considering the middle bent multimode waveguide being relatively long but having constant radius and width, interface I field is expanded into orthogonal eigenmodes in the bent waveguide and transmitted along region II by the EME method, and then goes to output port at region III calculated by the FDTD method for whole device transmission calculation. It can be seen from Fig. 5A that high transmission and high-quality self-image can be attained at certain location with the chosen Rmmi and θIO at 1561 nm, which can also be attained for four output center wavelengths around 1561 nm. Moreover, the length of the input / output tapers is selected to be LtIO=20μm to ensure greater than 99%transmission at 1561 nm, providing adiabatic transitions between waveguide widths WIO and Wsm, which can be shown in the EME simulation results in Fig. 5C. Further, Lymmi is chosen to be 120 μm to narrow the MMI waveguide width down to 150 nm, reducing the reflection of light remaining in the MMI waveguide after passing the output part.
[0061] F. Determination of Output Locations and Simulations of Transmission Spectra
[0062] The output locations of four output channels with center wavelength at 1591 nm, 1571 nm, 1551 nm, 1531 nm are defined as where the simulated transmission reaches maximum and highest-quality self-images are formed at their respective center wavelength. Thus, the locations of the output ports, which are defined as the arc lengths LIA, LIB, LIC, and LID as shown in Fig. 1A, are 842.03μm, 858.38μm, 875.27μm, and 892.56μm, respectively. The simulated spectra for four output ports are shown in Fig. 6A. The simulated insertion losses at four center wavelengths are 0.21 dB, 0.19 dB, 0.19 dB, 0.22 dB, respectively. Crosstalk (XT) is calculated to be smaller than -24.11dB at all center wavelengths. 3dB bandwidths (BW3dB) are greater than 12.1nm for all channels. Worst XT for 10nm bandwidth (XT10nm) is -16.9dB, -18.7dB, -20.0dB, -16.2dB, respectively. To illustrate the improved self-image quality of ABMMI, the spectra of the conventional AMMI are also simulated to compare as shown in Fig. 6B. Wmmi and WIO of the simulated AMMI are the same as ABMMI. θIO of the AMMI is optimized to be 15° to attain maximum transmission at 1561nm. Four AMMI’s output port locations are optimized to be LIA=885.3μm, LIB=901.6μm, LIC=918.4μm, and LID=935.7μm. It can be seen that due to higher phase error in the AMMI, center wavelength losses / XT have been degraded to 1.66dB / <-14.46dB. Also, worst XT10nm for all wavelength channels has been degraded to -8.98dB, -9.12 dB, -9.46dB and -11.72dB. As a consequence of higher self-image fidelity of the ABMMI, the self-image focuses better at the output waveguide, and the spectral side lobes are suppressed more effectively compared to the conventional AMMI, resulting in improved tolerance to laser wavelength drifting in CWDM applications. Additionally, it is noted that although the performance of the conventional AMMI may be further improved by increasing WIO and decreasing the portion of excited higher order modes with larger phase error [14, 15] , Wmmi would be required to be increased to keep enough spacing between output ports according to Equation (1) , which gives rise to a larger device footprint (MMI length is proportional to Wmmi2) and possibly higher IL. Moreover, LtIO would also need to be lengthened to keep adiabatic transition which leads to a larger device size. Nevertheless, the ABMMI of the subject invention ensures nearly all excited modes with low phase error by providing an extra design degree of freedom (Rmmi) and more flexible phase error optimization targets, whereas it could be hard to inhibit exciting modes with high phase error when limited to AMMI design.
[0063] Fig. 7 shows the calculated the channel central wavelength sensitivity with respect to the Wmmi. The calculation is conducted at 1551 nm channel output port of the ABMMI. It can be shown that a sensitivity of 184 pm / nm can be obtained, which is lower than AWG based filters [5, 6] , and filters based on normal CMZIs without specially designing the phase arms [8, 10] . It should be noted that the tolerance can be further improved by increasing Wmmi [14, 19] , while at the expense of a longer MMI and correspondingly larger insertion loss.
[0064] III. Fabrication and Measurement
[0065] The designed device was fabricated on the Applied Nanotools platform with 220nm thick silicon layer and 2.2μm / 2μm thick top / bottom oxide cladding as shown in Fig. 8. The silicon layer was patterned by electron beam lithography and inductively coupled plasma reactive-ion etching process, and the top cladding was deposited by plasma enhanced chemical vapor deposition process. The microscope image of the fabricated device is shown in Fig. 8. The on-chip device was coupled with fibers by grating couplers (GCs) working on C-band with TE modes. A pair of GCs connected by single mode waveguides were also fabricated near the designed device for spectra normalization. The spectra were measured with C-band tunable laser source and a photodetector. Light was route from laser to chip by single mode fibers (SMF) through a mechanical polarization controller (PC) to control the incident light being in TE state.
[0066] Fig. 9 shows the measured transmission spectra of the fabricated ABMMI from output ports O1-O4. The insertion losses at central wavelengths are 0.56 dB, 0.71 dB, 0.25 dB and 0.55 dB, respectively. XT at central wavelengths is 24.15 dB, 19.10 dB, 27.79 dB, 23.17 dB, respectively. For all wavelength channels the 3dB bandwidths are characterized to be BW3dB >11.7 nm. XT10nm is 14.48 dB, 17.90 dB, 13.76 dB, 11.16 dB for four channels, and the bandwidths for XT lower than -15 dB (BWXT<-15dB) are characterized as 10.9 nm, 12.6 nm, 11.4 nm and 12.1 nm, respectively. Agreement between experimental and simulation results can be observed despite the following discrepancies. Slightly higher XT (> -20 dB) for 1571 nm channel center is measured, which is probably because the non-perfect patterning of waveguide at output port O2 that make a small fraction of residual light in multimode waveguide output at the last port O2. Slightly higher IL than simulation may be attributed to surface roughness of the fabricated waveguide.
[0067] Discussion
[0068] Table I summarizes the performance comparison between the designed ABMMI and previously demonstrated AMMIs. First, compared to
[0013] , improved loss and XT performance are obtained, indicating lower self-imaging phase error is realized without utilizing shallow etched cladding which complicates the fabrication. Besides, it makes device more tolerant as the phase error and filter central wavelength could be more sensitive with respect to the shallow etching depth than Wmmi [17, 21] . Second, compared to [14-19] , the device of the subject invention has comparable and even better central wavelength XT without having to using a large Wmmi, the footprint is maintained to be relatively compatible. Third, by engineering the phase error with new degree of freedom Rmmi, the spectral side lobes of each output ports are much weaker and spectra shows more symmetry around central wavelength compared with other AMMI designs. This ensures enough low-XT BW which is an important indicator of CWDM applications. Lastly, though relatively higher central wavelength sensitivity is listed in Table 1, it should be noted that the sensitivity can also be improved if the ABMMI is designed with larger Wmmi, as in [14-15, 17-19] . Overall, the ABMMI has potential for CWDM filtering applications.
[0069] A C-band four channel CWDM filter based on ABMMI is provided. The performance of the ABMMI is ensured by improving the self-image fidelity of MMI. A novel degree of freedom, the radius of the bent MMI waveguide, and a relaxed phase error optimization target are introduced to enable low phase error. The ABMMI can be fabricated with single-step full etching and is demonstrated to be relatively compact compared with other AMMI based CWDM filters. The experiment results show low IL of 0.25 ~ 0.71 dB and low XT of -19.0 ~-27.8 dB at channel central wavelengths. Bandwidths for XT lower than -15 dB is 10.9 nm ~12.6 nm for all channels.
[0070] In one embodiment, the ABMMI can also be adapted and implemented on platforms with relatively lower index contrast, such as lithium niobate on insulator (LNOI) or silicon nitride (SiN) , to reduce phase error, with a device footprint larger than a Si-based design for the same functionality of CWDM4 with 20 nm wavelength spacing. Additionally, the ABMMI can be designed for WDM applications with more wavelength channels and finer wavelength spacing by employing a larger Wmmi which is the width of the multimode waveguide, though this comes at the expense of an increased device footprint.
[0071] All patents, patent applications, provisional applications, and publications referred to or cited herein are incorporated by reference in their entirety, including all figures and tables, to the extent they are not inconsistent with the explicit teachings of this specification.
[0072] It should be understood that the examples and embodiments described herein are for illustrative purposes only and that various modifications or changes in light thereof will be suggested to persons skilled in the art and are to be included within the spirit and purview of this application. In addition, any elements or limitations of any invention or embodiment thereof disclosed herein can be combined with any and / or all other elements or limitations (individually or in any combination) or any other invention or embodiment thereof disclosed herein, and all such combinations are contemplated with the scope of the invention without limitation thereto.
[0073] EXEMPLARY EMBODIMENTS
[0074] Embodiment 1. A Coarse Wavelength Division Multiplexing (CWDM) filter system based on an angled bent multimode interferometer (ABMMI) , comprising:
[0075] one input port;
[0076] a plurality of output ports; and
[0077] a bent multimode waveguide connecting the input port and the plurality of output ports, having an inner radius of Rmmi;
[0078] wherein light incident to the input port is demultiplexed into a plurality of wavelength channels.
[0079] Embodiment 2. The CWDM filter system of embodiment 1, wherein each of the input port and plurality of output ports comprises a first waveguide having a width WIOand a single mode waveguide having a width Wsm connected to the first waveguide, wherein the width WIO is greater than the width Wsm.
[0080] Embodiment 3. The CWDM filter system of embodiment 1, wherein the plurality of output ports are four output ports.
[0081] Embodiment 4. The CWDM filter system of embodiment 1, wherein each of the plurality of output ports corresponds to a channel and the demultiplexed light is output from the plurality of channels, respectively.
[0082] Embodiment 5. The CWDM filter system of embodiment 3, wherein the incident light demultiplexed into the plurality of wavelength channels has center wavelengths of 1591nm, 1571nm, 1551nm and 1531nm, respectively.
[0083] Embodiment 6. The CWDM filter system of embodiment 1, wherein the inner radius Rmii is optimized to form a high-quality self-image at the output ports.
[0084] Embodiment 7. The CWDM filter system of embodiment 1, wherein an excited mode order range of the CWDM filter system is optimized to form a high-quality self-image at the output ports.
[0085] Embodiment 8. The CWDM filter system of embodiment 2, wherein an angle between a middle portion of respective input port and the output portions and tangent of an inner arc of the bent multimode waveguide is defined as θIO.
[0086] Embodiment 9. The CWDM filter system of embodiment 8, wherein θIO and WIO are optimized for sufficient distances between adjacent output ports for channel spacing such that a minimal crosstalk between output wavelength channels is obtained.
[0087] Embodiment 10. The CWDM filter system of embodiment 1, wherein Insertion Loss (IL) in a range between 0.25 dB and 0.71 dB and Crosstalk (XT) in a range between -19.0 dB and -27.8 dB at channel central wavelengths are obtained.
[0088] Embodiment 11. A method for optimizing parameters of a Coarse Wavelength Division Multiplexing (CWDM) filter system based on angled bent multimode interferometer (ABMMI) comprising one input port; a plurality of output ports; and a bent multimode waveguide connecting the input port and the plurality of output ports, having an inner radius of Rmmi; wherein light incident to the input port is demultiplexed into a plurality of wavelength channels, the method comprising:
[0089] optimizing the inner radius Rmmi to form a high-quality self-image at the output ports.
[0090] Embodiment 12. The method of embodiment 11, further comprising optimizing an excited mode order range of the CWDM filter system to form a high-quality self-image at the output ports.
[0091] Embodiment 13. The method of embodiment 12, wherein the input port and the output ports each has a width WIO and an angle between a middle portion of respective input / output and tangent of an inner arc of the bent multimode waveguide is defined as θIO.
[0092] Embodiment 14. The method of embodiment 13, further comprising optimizing θIO and WIO for sufficient distances between adjacent output ports for channel spacing such that a minimal crosstalk between output wavelength channels is obtained.
[0093] Embodiment 15. The method of embodiment 11, wherein Insertion Loss (IL) in a range between 0.25 dB and 0.71 dB and Crosstalk (XT) in a range between -19.0 dB and -27.8 dB at channel central wavelengths are obtained.
[0094] Embodiment 16. A non-transitory computer readable medium having stored therein program instructions executable by a computing system to cause the computing system to perform a method for optimizing parameters of a Coarse Wavelength Division Multiplexing (CWDM) filter system based on angled bent multimode interferometer (ABMMI) comprising one input port; a plurality of output ports; and a bent multimode waveguide connecting the input port and the plurality of output ports, having an inner radius of Rmmi; wherein light incident to the input port is demultiplexed into a plurality of wavelength channels, the method comprising:
[0095] optimizing the inner radius Rmmi to form a high-quality self-image at the output ports.
[0096] Embodiment 17. The non-transitory computer readable medium of embodiment 16, further comprising optimizing an excited mode order range of the CWDM filter system to form a high-quality self-image at the output ports.
[0097] Embodiment 18. The non-transitory computer readable medium of embodiment 16, wherein the input port and the output ports each has a width WIO and an angle between a middle portion of respective input port and output ports and tangent of an inner arc of the bent multimode waveguide is defined as θIO.
[0098] Embodiment 19. The non-transitory computer readable medium of embodiment 18, further comprising optimizing θIO and WIO for sufficient distances between adjacent output ports for channel spacing such that a minimal crosstalk between output wavelength channels is obtained.
[0099] Embodiment 20. The non-transitory computer readable medium of embodiment 16, wherein Insertion Loss (IL) in a range between 0.25 dB and 0.71 dB and Crosstalk (XT) in a range between -19.0 dB and -27.8 dB at channel central wavelengths are obtained.
[0100] Embodiment 21. The CWDM filter system of embodiment 1, wherein the bent multimode waveguide comprises a layer of silicon and a layer of silicon dioxide disposed on the layer of silicon.
[0101] Embodiment 22. The CWDM filter system of embodiment 1, wherein the bent multimode waveguide is made of a lithium niobate on insulator (LNOI) .
[0102] Embodiment 23. The CWDM filter system of embodiment 1, wherein the bent multimode waveguide is made of silicon nitride (SiN) .
[0103] REFERENCES
[0104] [1] L. Liao, S. Fathololoumi, K. Nguyen, H. Mahalingam, D. Hui, J. Heck, H. Frish, R. Defrees, C. Malouin, P. Seddighian, M. Huang, K. Al-Hemyari, Y.J. Chen, Y. Wang, W. Lin, D. Zhu, R. Jones, Y. Akulova, and T. Liljeberg, "Silicon Photonics for Next-Generation Optical Connectivity, " in 2023 Optical Fiber Communications Conference and Exhibition (OFC) , 2023) , 1-3.
[0105] [2] N. Jouppi, G. Kurian, S. Li, P. Ma, R. Nagarajan, L. Nai, N. Patil, S. Subramanian, A. Swing, B. Towles, C. Young, X. Zhou, Z. Zhou, and D. A. Patterson, "TPU v4: An Optically Reconfigurable Supercomputer for Machine Learning with Hardware Support for Embeddings, " in Proceedings of the 50th Annual International Symposium on Computer Architecture, (Association for Computing Machinery, Orlando, FL, USA, 2023) , p. Article 82.
[0106] [3] “Spectral grids for WDM applications: CWDM wavelength grid, ” ITU-T Recommendation G. 694.2 (2003) .
[0107] [4] W. Shi, H. Yun, C. Lin, M. Greenberg, X. Wang, Y. Wang, S.T. Fard, J. Flueckiger, N.A.F. Jaeger, and L. Chrostowski, "Ultra-compact, flat-top demultiplexer using anti-reflection contra-directional couplers for CWDM networks on silicon, " Optics Express 21, 6733-6738 (2013) .
[0108] [5] 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 Journal 6, 1-9 (2014) .
[0109] [6] S.K. Selvaraja, W. Bogaerts, P. Dumon, D. Van Thourhout, and R. Baets, "Subnanometer Linewidth Uniformity in Silicon Nanophotonic Waveguide Devices Using CMOS Fabrication Technology, " IEEE Journal of Selected Topics in Quantum Electronics 16, 316-324 (2010) .
[0110] [7] D. Liu, M. Zhang, Y. Shi, and D. Dai, "Four-Channel CWDM (de) Multiplexers Using Cascaded Multimode Waveguide Gratings, " IEEE Photonics Technology Letters 32, 192-195 (2020) .
[0111] [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, " Optics Express 21, 11652-11658 (2013) .
[0112] [9] H. Xu, L. Liu, and Y. Shi, "Polarization-insensitive four-channel coarse wavelength-division (de) multiplexer based on Mach-Zehnder interferometers with bent directional couplers and polarization rotators, " Opt Lett 43, 1483-1486 (2018) .
[0113]
[0010] H. Xu and Y. Shi, "Flat-Top CWDM (De) Multiplexer Based on MZI With Bent Directional Couplers, " IEEE Photonics Technology Letters 30, 169-172 (2018) .
[0114]
[0011] T. -H. Yen and Y. Hung, Jr., "Fabrication-Tolerant CWDM (de) Multiplexer Based on Cascaded Mach–Zehnder Interferometers on Silicon-on-Insulator, " Journal of Lightwave Technology 39, 146-153 (2021) .
[0115]
[0012] H. Xu, D. Dai, and Y. Shi, "Low-crosstalk and fabrication-tolerant four-channel CWDM filter based on dispersion-engineered Mach-Zehnder interferometers, " Optics Express 29, 20617-20631 (2021) .
[0116]
[0013] Y. Hu, R.M. Jenkins, F.Y. Gardes, E.D. Finlayson, G.Z. Mashanovich, and G.T. Reed, "Wavelength division (de) multiplexing based on dispersive self-imaging, " Opt Lett 36, 4488-4490 (2011) .
[0117]
[0014] Y. Hu, F.Y. Gardes, D.J. Thomson, G.Z. Mashanovich, and G.T. Reed, "Coarse wavelength division (de) multiplexer using an interleaved angled multimode interferometer structure, " Applied Physics Letters 102, 251116 (2013) .
[0118]
[0015] Z. Zhang, Q. Cheng, K. Zhang, H. Zhang, Z. Zhang, B. Huang, H. Liu, H. Li, P. Niu, and H. Chen, "Vertical Fibre Interfacing Interleaved Angled MMI for Thermal-Tuning-Free Wavelength Division (de) Multiplexing and Low-Cost Fibre Packaging, " Journal of Lightwave Technology 39, 6260-6268 (2021) .
[0119]
[0016] Y. Hu, T. Li, D.J. Thomson, X. Chen, J.S. Penades, A.Z. Khokhar, C.J. Mitchell, G.T. Reed, and G.Z. Mashanovich, "Mid-infrared wavelength division (de) multiplexer using an interleaved angled multimode interferometer on the silicon-on-insulator platform, " Opt Lett 39, 1406-1409 (2014) .
[0120]
[0017] G. Chen, Z. Ruan, Z. Wang, P. Huang, C. Guo, D. Dai, K. Chen, and L. Liu, "Four-channel CWDM device on a thin-film lithium niobate platform using an angled multimode interferometer structure, " Photon. Res. 10, 8-13 (2022) .
[0121]
[0018] T.D. Bucio, A.Z. Khokhar, G.Z. Mashanovich, and F.Y. Gardes, "N-rich silicon nitride angled MMI for coarse wavelength division (de) multiplexing in the O-band, " Opt Lett 43, 1251-1254 (2018) .
[0122]
[0019] T.D. Bucio, A.Z. Khokhar, G.Z. Mashanovich, and F.Y. Gardes, "Athermal silicon nitride angled MMI wavelength division (de) multiplexers for the near-infrared, " Optics Express 25, 27310-27320 (2017) .
[0123]
[0020] L.B. Soldano and E.C.M. Pennings, "Optical multi-mode interference devices based on self-imaging: principles and applications, " Journal of Lightwave Technology 13, 615-627 (1995) .
[0124]
[0021] C. Ye, M. Zhang, Y. Shi, and D. Dai, "Broadband dual-mode 2 ×2 3 dB multimode interference couplers with a shallowly etched multimode section, " Appl Optics 59, 7308-7312 (2020) .
[0125]
[0022] R. Halir, G. Roelkens, A. and J.G. Wangüemert-Pérez, "High-performance 90° hybrid based on a silicon-on-insulator multimode interference coupler, " Opt Lett 36, 178-180 (2011) .
[0126]
[0023] J.Z. Huang, R. Scarmozzino, and R.M. Osgood, "A new design approach to large input / output number multimode interference couplers and its application to low-crosstalk WDM routers," IEEE Photonics Technology Letters 10, 1292-1294 (1998) .
Claims
1.A Coarse Wavelength Division Multiplexing (CWDM) filter system based on an angled bent multimode interferometer (ABMMI) , comprising:one input port;a plurality of output ports; anda bent multimode waveguide connecting the input port and the plurality of output ports, having an inner radius of Rmmi;wherein light incident to the input port is demultiplexed into a plurality of wavelength channels.2.The CWDM filter system of claim 1, wherein each of the input port and plurality of output ports comprises a first waveguide having a width WIO and a single mode waveguide having a width Wsm connected to the first waveguide, wherein the width WIO is greater than the width Wsm.3.The CWDM filter system of claim 1, wherein the plurality of output ports are four output ports.4.The CWDM filter system of claim 1, wherein each of the plurality of output ports corresponds to a channel and the demultiplexed light is output from the plurality of channels, respectively.5.The CWDM filter system of claim 3, wherein the incident light demultiplexed into the plurality of wavelength channels has center wavelengths of 1591nm, 1571nm, 1551nm and 1531nm, respectively.6.The CWDM filter system of claim 1, wherein the inner radius Rmmi is optimized to form a high-quality self-image at the output ports.7.The CWDM filter system of claim 1, wherein an excited mode order range of the CWDM filter system is optimized to form a high-quality self-image at the output ports.8.The CWDM filter system of claim 2, wherein an angle between a middle portion of respective input port and the output portions and tangent of an inner arc of the bent multimode waveguide is defined as θIO.9.The CWDM filter system of claim 8, wherein θIO and WIO are optimized for sufficient distances between adjacent output ports for channel spacing such that a minimal crosstalk between output wavelength channels is obtained.10.The CWDM filter system of claim 1, wherein Insertion Loss (IL) in a range between 0.25 dB and 0.71 dB and Crosstalk (XT) in a range between -19.0 dB and -27.8 dB at channel central wavelengths are obtained.11.A method for optimizing parameters of a Coarse Wavelength Division Multiplexing (CWDM) filter system based on angled bent multimode interferometer (ABMMI) comprising one input port; a plurality of output ports; and a bent multimode waveguide connecting the input port and the plurality of output ports, having an inner radius of Rmmi; wherein light incident to the input port is demultiplexed into a plurality of wavelength channels, the method comprising:optimizing the inner radius Rmmi to form a high-quality self-image at the output ports.12.The method of claim 11, further comprising optimizing an excited mode order range of the CWDM filter system to form a high-quality self-image at the output ports.13.The method of claim 12, wherein the input port and the output ports each has a width WIO and an angle between a middle portion of respective input / output and tangent of an inner arc of the bent multimode waveguide is defined as θIO.14.The method of claim 13, further comprising optimizing θIO and WIO for sufficient distances between adjacent output ports for channel spacing such that a minimal crosstalk between output wavelength channels is obtained.15.The method of claim 11, wherein Insertion Loss (IL) in a range between 0.25 dB and 0.71 dB and Crosstalk (XT) in a range between -19.0 dB and -27.8 dB at channel central wavelengths are obtained.16.A non-transitory computer readable medium having stored therein program instructions executable by a computing system to cause the computing system to perform a method for optimizing parameters of a Coarse Wavelength Division Multiplexing (CWDM) filter system based on angled bent multimode interferometer (ABMMI) comprising one input port; a plurality of output ports; and a bent multimode waveguide connecting the input port and the plurality of output ports, having an inner radius of Rmmi; wherein light incident to the input port is demultiplexed into a plurality of wavelength channels, the method comprising:optimizing the inner radius Rmmi to form a high-quality self-image at the output ports.17.The non-transitory computer readable medium of claim 16, further comprising optimizing an excited mode order range of the CWDM filter system to form a high-quality self-image at the output ports.18.The non-transitory computer readable medium of claim 16, wherein the input port and the output ports each has a width WIO and an angle between a middle portion of respective input port and output ports and tangent of an inner arc of the bent multimode waveguide is defined as θIO.19.The non-transitory computer readable medium of claim 18, further comprising optimizing θIO and WIO for sufficient distances between adjacent output ports for channel spacing such that a minimal crosstalk between output wavelength channels is obtained.20.The non-transitory computer readable medium of claim 16, wherein Insertion Loss (IL) in a range between 0.25 dB and 0.71 dB and Crosstalk (XT) in a range between -19.0 dB and -27.8 dB at channel central wavelengths are obtained.