Single-mode polarization insensitive grating coupler

Abstract

We provide a general two dimensional grating coupler system that maximizes its coupling power, while simultaneously minimizing the polarization dependent loss ( PDL) for any linearly polarized input. This approach solves the long lasting problem of complete polarization insensitive fiber to chip grating coupler designs, using a passive device with no active control. In other embodiments, an arbitrary elliptical input state of polarization is dealt with using a passive scattering structure having two waveguide outputs into which the input free space beam is split according to its polarization components. A simple active control scheme is then used to combine these two waveguide outputs in phase to provide a single output with minimal PDL for arbitrary input state of polarization.

Description

[0001] Single-Mode Polarization Insensitive Grating

[0002] Coupler

[0003] FIELD OF THE INVENTION

[0004] This invention relates to coupling of a free space optical mode to a waveguide mode .

[0005] BACKGROUND

[0006] Coupling light between an optical fiber and an on-chip optical waveguide is a common practical problem, e.g., for fiber coupling of an integrated photonic circuit. The main practical issue is that standard optical fiber provides a random state of polarization at its output, while the on- chip waveguide has a well-defined state of polarization. This problem originates from the inherent characteristics of standard single-mode fiber (non-polarization-maintaining fiber) having two degenerate polarization modes basis (s and p- polarizations when the plane of incidence is defined) , and thus its output state of polarization typically being both arbitrary and time-varying. Because most grating couplers used to couple power from fiber are input polarization dependent, the coupling power varies due to the polarization-dependent loss (PDL) . Thus, there is a need to account for this polarization issue when mode matching the fiber to the waveguide.

[0007] Conventional ways of dealing with this issue suffer from various disadvantages. For example, the use of polarization-maintaining fiber adds cost and complicates integration with the rest of the optical system, since use of polarization-maintaining fiber throughout an entire system is usually not an option because polarization maintaining fibers are much more expensive than single mode fibers .

[0008] Another approach is a polarization diversity scheme, which means integrating two separate photonic components dedicated to s and p polarization inputs. Such functionality can typically be achieved by the combination of polarization splitting grating, polarization splitter and rotator. However, this requires integrating duplicate photonic components for each polarization which leads to larger footprint of the overall design.

[0009] SUMMARY

[0010] The present work provides passive couplers and active polarization compensation techniques that are truly polarization-insensitive, meaning that arbitrary polarization of the fiber output always couples to a single mode of the waveguide, e.g., a TE mode, with nearly the same efficiency across different linear polarizations in the fiber. This means that regardless of s-polarization or p- polarization of the fiber, the grating coupler can couple into a single waveguide mode, such as the TE mode. This way, it does not matter what the input state of polarization from the fiber is (or how it changes over time) . The output is always polarization matched to the on-chip waveguide. This contrasts with "polarization insensitive grating couplers" as usually considered in the literature, which refer to couplers that couple the orthogonal linear polarizations of the fiber, s-polarization and p-polarization, into different waveguide modes, TE and TM modes, respectively, with similar coupling efficiency.

[0011] For the passive operation, we resolve this problem by designing a general two dimensional grating coupler system that maximizes its coupling power, while simultaneously minimizing the polarization dependent loss (PDL) for linear polarization input as the figure of merit of the device. This approach solves the long lasting problem of complete polarization insensitive fiber to chip grating coupler designs, using a passive device with no active control.

[0012] For example, photonic inverse design can directly target the combined metrics of minimizing insertion loss and minimizing PDL. We can design a passive grating coupler element that passively couples equal amounts of final coupled power from both linear polarization basis of the fiber output mode, therefore covering arbitrary linear polarization of the output. This allows us to no longer rely on active phase tuning element for coherent combination and we can passively achieve low PDL.

[0013] For the active polarization scheme, we developed a general feedforward compensation scheme. Such tasks can be achieved by splitting the s- and p- polarization of the input with polarization splitting grating couplers (PSGC) and combining into a single mode output waveguide. The control system detects the relative phase of s- and p- polarization input and tunes the combining path based on the detection result to eliminate PDL for arbitrary input polarization .

[0014] Photonic technology in general can benefit from this technology as it will simplify the system requirements, and no longer require polarization maintaining fibers and / or larger footprint for polarization diversity. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] FIG. 1 schematically shows the operating principle of some embodiments of the invention.

[0016] FIG. 2 is an SEM (scanning electron microscope) image of an exemplary fabricated device.

[0017] FIG. 3 schematically shows the difference between a polarization insensitive grating coupler and a conventional grating coupler.

[0018] FIG. 4 shows simulated and experimental coupling efficiency results.

[0019] FIG. 5 shows a comparison of experimental PDL results between a conventional grating coupler and a coupler designed according to an embodiment of the invention.

[0020] FIG. 6 schematically shows the operating principle of some embodiments of the invention.

[0021] FIG. 7 shows an example of the performance of the approach of FIG. 6.

[0022] DETAILED DESCRIPTION

[0023] FIG. 1 schematically shows the operating principle of some embodiments of the invention. Light from free space mode 108 (at a prescribed angle of incidence) is incident on the part of device 102 that is indicated by 104a and 104b, and is coupled to waveguide mode 110 of waveguide 106. Waveguide mode 110 has a defined state of polarization (typically transverse electric (TE) ) , so what is contemplated here is true passive polarization insensitivity as described above. Here 102 is an inverse-designed passive scattering structure that is numerically optimized to minimize a cost function. This cost function penalizes:

[0024] Li, a first loss from a first polarization of the free space mode 108 to the waveguide mode 110;

[0025] L2, a second loss from a second polarization of the free space mode 108 to the waveguide mode 110 (typically the second polarization is taken to be orthogonal to the first polarization) ; and polarization dependent loss (PDL) . Here PDL is defined as the ratio of maximum coupling efficiency to minimum coupling efficiency over all input states of polarization.

[0026] An exemplary cost function f for Li, L2 and PDL is given by where ci and C2 are weights to set the relative importance of minimizing insertion loss and minimizing PDL. The numerical optimization is to minimize f by adjusting the scattering pattern in device 102. In principle there is no restriction on the resulting lateral pattern. In practice constraints may be placed on the lateral pattern to account for fabrication limitations. As is well known in the field of numerical optimization, the specific form of the cost function is not crucial, provided it penalizes both insertion loss and PDL.

[0027] The resulting passive planar scattering structure is schematically shown as 102 on FIG. 1, and is not amenable to a description of its structure.

[0028] FIG. 2 is an SEM (scanning electron microscope) image of an exemplary fabricated device. FIG. 3 schematically shows the difference between a polarization insensitive grating coupler and a conventional grating coupler.

[0029] FIG. 4 shows simulated and experimental coupling efficiency results for a polarization-insensitive couple as on FIG. 1.

[0030] FIG. 5 shows a comparison of experimental PDL results between a conventional grating coupler and a coupler designed according to an embodiment of the invention. These results should about 30 dB reduction in PDL.

[0031] The preceding work using passive structures effectively assumes the incident state of polarization is an arbitrary linear state of polarization (i.e. the two polarization components are in phase) . In general there can be a phase difference between these two polarization components, resulting in elliptical (or circular) polarization. The preceding work is not directly applicable to this more general case, which motivates the following embodiment.

[0032] FIG. 6 schematically shows the operating principle of some embodiments of the invention. Here 602 is a polarization-splitting grating coupler (PSGC) , 604 is a phase detection subsystem, 606 is a heater, 608 is a 2 to 1 coupler (e.g., a multi-mode interference (MMI) device) , 610 is a compensated single-mode output, and 612 is a controller. This control scheme detects the relative phase of s- and p- polarization input and tunes the combining path via heater 606 based on the detection result to eliminate PDL for arbitrary input polarization by ensuring a 90-degree phase difference at the inputs of coupler 608. This can be done in an analog way with simple lock-in detection within phase detection block 604, thereby advantageously avoiding gradient descent and other complicated computational approaches often used in conventional active polarization combiners. A calibration may be needed to relate the measured phase in block 604 to the input phase difference at coupler 608, but the need for such calibration may be reduced or eliminate in preferred embodiments where path length differences and the like are eliminated in a fully integrated design. Although the results below are obtained with this feedforward scheme, practice of the invention does not depend critically on how the two inputs to coupler 608 are made to differ in phase by 90 degrees. Any known feedforward or feedback approach can be used to perform this function .

[0033] This feedforward scheme can be regarded as a phase-only compensation method, which has the advantage of being simpler than conventional polarization compensation methods that adjust both phase and amplitude. However, phase-only compensation has the disadvantage of having a maximum efficiency of 50%.

[0034] Here PSGC 602 is preferably an inverse-designed passive polarization component analogous to 102 on FIG. 1, but it is based on minimization of a different cost function. The main advantage of such an inverse-designed PSGC is the ability to design it for broadband performance. In particular, PSGC 602 has a free space input and two waveguide outputs Oi and O2 having the same polarization (e.g., both transverse electric (TE) or both transverse magnetic (TM) ) . The cost function here penalizes:

[0035] Li, a first loss from a first polarization component of the free space mode 108 to Oi; and

[0036] L2, a second loss from a second polarization of the free space mode 108 to O2. No attempt to mitigate PDL is made in this cost function. Instead, in this example, that function is performed by the control scheme of FIG. 6. In some cases, symmetry can be exploited to reduce the cost function to a simpler form that is equivalent to the above-described cost function. For example, in a structure where symmetry forces the same coupling efficiency for s-polarization into a first waveguide as for p-polarization into a second waveguide, the cost function can simply penalize the coupling loss for s- polarization into the first waveguide.

[0037] FIG. 7 shows an example of the performance of the approach of FIG. 6.

Claims

CLAIMS1. A method of making a photonic component, the method comprising : performing numerical optimization of a passive planar photonic technology scattering structure; wherein the passive planar photonic technology scattering structure is configured to couple two orthogonal polarization components of an incident linearly polarized free space mode to a single polarization of a single waveguide mode; wherein a cost function of the numerical optimization penalizes : a) a first loss from a first polarization component of the free space mode to the waveguide mode; b) a second loss from a second polarization component of the free space mode to the waveguide mode; and c) polarization dependent loss.

2. The method of claim 1, wherein the free space mode is provided by radiation from an end face of an optical fiber.

3. The method of claim 1, wherein the waveguide mode is a transverse electric (TE) waveguide mode.

4. A method of making a photonic component, the method comprising : performing numerical optimization of a passive planar photonic technology scattering structure;wherein the passive planar photonic technology scattering structure is configured to couple two orthogonal polari zation components of an incident free space mode to a first waveguide output and to a second waveguide output ; wherein a cost function of the numerical optimi zation penali zes : a ) a first loss from a first polari zation component of the free space mode to the first waveguide output ; and b ) a second loss from a second polari zation component of the free space mode to the second waveguide output .5 . The method of claim 4 , wherein the free space mode is provided by radiation from an end face of an optical fiber .6 . The method of claim 1 , wherein waveguide modes for both waveguide outputs are transverse electric ( TE ) waveguide modes .7 . A method of making an active free-space to waveguide coupler, the method comprising : making a photonic component according to the method of claim 4 ; providing a control system configured to combine the first waveguide output and the second waveguide output with a 90-degree phase di f ference into a single output waveguide mode .

Images