CN100362394C - Integrated Optical Polarization Converter Based on Helical Properties of Spatial Structure - Google Patents

Abstract

An integrated optical polarization converter comprising: a plurality of core layers for simulating a gradually twisted waveguide in which a propagating mode is adiabatically transformed from an initial polarization state to a different final polarization state.

Description

Integrated Optical Polarization Converter Based on Helical Properties of Spatial Structure

technical field

The invention relates to the field of integrated optical polarization converters, in particular to integrated optical polarization converters based on mode evolution or spatial structure helical properties.

Background technique

As the popularity of optical fiber communications grows, the demand for more refined processing of optical signals continues to grow. Since integrated optics devices can integrate many optical functions on a single chip, integrated optics approaches have the potential to meet the requirements for more refined optical signal processing. However, in order to increase functionality and reduce cost per function, the density of components on a chip must be increased.

At a given wavelength, mode confinement in a dielectric waveguide is determined by the contrast between the refractive index of the core and the cladding. The higher the contrast, the tighter the mode restriction. The result of the tight confinement enables denser waveguides and directs light around acute angles without significant radiative losses. Since these are the two most important parameters affecting device density, it can roughly be said that the higher the index contrast, the higher the device density. However, as the refractive index contrast increases, the transverse electric (TE) mode and transverse magnetic (TM) mode propagating in the waveguide start to exhibit different characteristics. In the straight part of the square waveguide, the TE mode and the TM mode propagate at the same speed, while in the curved part, the TE mode and the TM mode propagate at very different speeds. Furthermore, when coupling a pair of square high-index-contrast (HIC) waveguides, TE and TM modes tend to couple at different speeds. Since most integrated optical components are sensitive to propagation velocity and waveguide-to-waveguide coupling, these effects lead to polarization-dependent performance, a consequence that is incompatible with firing random polarization states from standard single-mode fibers in telecommunication applications.

One way to compensate for these effects is to use rectangular waveguide structures and vary the aspect ratio of the waveguides to compensate for the natural differences in propagation around acute angles and/or to equalize waveguide-to-waveguide coupling. However, while it is possible to compensate for one of these effects for a particular device in this way, it becomes difficult or impossible to compensate for all devices on a chip simultaneously as the index contrast increases.

Another way to overcome the polarization sensitivity in the HIC integrated optical circuit is to split the random input polarized light emitted from the single-mode (SM) fiber by using a polarization beam splitter (PBS), and couple the output light to the polarization-maintaining (PM) fiber, One of these PM fibers is twisted 90°, and the two fibers are coupled to separate paths on the IC chip. On each of these paths, the same structure is used to process the two components independently. At the output, these components are recombined by coupling to another pair of PM fibers, twisting the PM fiber on the previously untwisted path, and coupling these two fibers to another PBS with an SM fiber output. This approach is often called a "polarization diversity" scheme, and while it is feasible with bulk optics, it is cumbersome. Aligning PM fibers is difficult and expensive. Furthermore, to preserve signal integrity, the path lengths must be matched to within at least 1/10 of the bit length (i.e., approximately 2mm for a 10Gb/s signal and approximately 0.5mm for a 40Gb/s signal, assuming a refractive index of 1.5 ).

A better method is to integrate the beam splitting function of the PBS and the rotation function of the twisted PM fiber into an integrated optical chip. Doing so can eliminate the need to align the PM fibers, and optical path lengths can be easily matched using photolithography.

Several integrated optical polarizing beam splitters and rotators (or converters) have been proposed. However, most devices proposed to date rely on the coupling of a pair of waveguide modes. Devices based on coupled modes generally have wavelength sensitivity due to differences in the dispersion of the supermodes propagating in the waveguide structure. Furthermore, this method is very sensitive to machining errors. Even small changes in waveguide geometry or spacing can have a significant impact on device performance.

A better way to make polarizing beam splitters or rotators is to use the principle of mode evolution. By making gradual (or adiabatic) changes to the waveguide geometry, it is possible to tune the modes in the waveguide and to separate or rotate the polarization states. This approach only requires that no power is exchanged between the modes, which can be ensured by proper design of the waveguide and slow evolution of the structure. Since preventing mode coupling is a relatively relaxed requirement, devices based on mode evolution tend to be wavelength insensitive and allow for fabrication errors. It has been proposed and demonstrated that polarization beam splitters based on mode evolution can be fabricated. However, this method has the disadvantage of requiring multiple waveguide materials. In addition, no polarization converters using the principle of mode evolution have been proposed so far.

In general, it is an object of the present invention to convert TM input polarized light to TE output polarized light or vice versa using integrated optics based on mode evolution or slructural chirality.

It is another object of the invention that such devices are wavelength insensitive, tolerate fabrication errors, and require only a single material system to be constructed.

These and other objects of the present invention will become apparent to those skilled in the art from the following detailed description and accompanying drawings.

Contents of the invention

According to one feature of the invention, an integrated optical polarization converter is provided. The integrated optical polarization converter includes multiple core layers to simulate a gradually twisted waveguide in which a propagating mode is adiabatically transformed from an initial polarization state to a different final polarization state.

According to another feature of the invention, a method of forming an integrated optical polarization converter is provided. The method includes receiving an initial state of polarization. Additionally, the method includes forming a plurality of core layers for simulating a gradually twisted waveguide, and wherein a propagating mode is adiabatically transformed from an initial polarization state to a different final polarization state.

Description of drawings

Figure 1 is a schematic diagram of a twisted waveguide;

Fig. 2 is a schematic diagram of a polarization converter utilizing three adiabatic thinning core layers;

3A-3B are gray scale images of the fundamental mode electric field propagating in the waveguide of the present invention;

Fig. 4 is the function graph of the performance of the device in concrete implementation Fig. 2 as this device length;

Fig. 5 is the function graph of the performance of the device in concrete implementation Fig. 2 as electromagnetic field wavelength;

Figure 6 is a schematic diagram of a polarization converter utilizing three adiabatically tapered and separated core layers;

Fig. 7 is a schematic diagram of a polarization converter using heat insulation to thin the middle layer and heat insulation to separate the upper and lower core layers;

Figure 8 is a graph showing the performance of the device in Figure 7 as a function of the length of the device;

Fig. 9 is a function graph of the performance of the device as the wavelength of the electromagnetic field in the specific implementation of Fig. 7;

Figure 10 is a schematic diagram of a polarization converter utilizing only two core layers that are adiabatically tapered and separated;

Figure 11 is a graph of the performance of the device of an embodiment of Figure 10 as a function of the length of the device; and

Figure 12 is a graph of the performance of the device of an embodiment of Figure 10 as a function of electromagnetic field wavelength.

Detailed ways

The mode structure of a general rectangular dielectric waveguide is composed of the infinite sum of the smallest two guided electromagnetic modes, namely, TE (or quasi-TE) mode and TM (or quasi-TM) mode, and unguided (or radiating) electromagnetic modes. If the rectangular waveguide is rotated by 90°, its mode structure is similarly rotated, where TE modes become TM modes and TM modes become TE modes. Therefore, a smooth transition between the rectangular waveguide and its rotating waveguide should enable polarization conversion through mode evolution. However, some perturbations to the initial structure can induce coupling between modes. For the mode evolution method to be effective, power swapping between modes must be prohibited.

One way to make the transition between a rectangular waveguide and its rotated counterpart is to twist its initial structure. FIG. 1 is a schematic diagram of an adiabatic twisted dielectric waveguide 100 having an input 102 and an output 104 . The function of the twisted waveguide 100 is to perturb the mode structure of the rectangular waveguide, thereby inducing the coupling between each guided mode and between the guided mode and the radiation mode. Since the coupling between the guided and radiating modes requires strong perturbations, the coupling to the radiating modes is usually negligible in slowly evolving structures. However, the coupling between guided modes still remains an important influence.

If the waveguide 100 in Figure 1 is square, the guided modes are degenerate, so they propagate at the same speed. The twist-induced field coupling superimposes coherently along the length of the structure, and the power exchange between individual modes is significant. This is a redundant result because the twisted waveguide 100 works on the principle of mode evolution rather than mode coupling. To prevent coupling, rectangular waveguides with large aspect ratios are utilized, so the guided modes propagate at different velocities. In this case, the power coupled from one mode to another is incoherently added along the length of the structure, as long as the structure is long enough to phase shift the modes. As the index contrast, aspect ratio, and transition length increase, so does the degree of incoherence, and the accumulated power exchange along the length of the structure can be made arbitrarily low. Its performance deviates from ideal only if the transition length becomes too short or the waveguide aspect ratio is too small for phase shifting to occur.

The structure 100 in Figure 1 is an ideal structure in which the waveguide is fully twisted. However, in a real device, all such geometries should be fabricated using micromachining techniques, which usually require layering processes to fabricate such structures, whose features are defined by photolithography. Therefore, it is ideal to model such twisted waveguides with a finite number of layers. Here, a layer is defined as a horizontal slice through the waveguide cross-section, which has no refractive index variation along the vertical direction.

Optical waveguides are usually made of dielectric materials with different refractive indices. Generally speaking, a material with a higher refractive index is used as a core material, and a material with a lower refractive index is used as a cladding material. Specifically, the cladding material is defined as the material with the lowest refractive index within the layer. Therefore, all other materials within this layer are core material. A core layer is defined as a layer comprising a core material.

The basic requirements for the structure of the polarization converter are quite loose, and the main requirement is to preserve the helical characteristic of the spatial structure, or the twist of the electromagnetic field direction. Several possible geometries are described below.

Fig. 2 is a schematic diagram of an integrated optical polarization converter 2 using three core layers 4, 6 and 8, the heights of which are h ₁ , h ₂ and h ₃ The aligned rectangular waveguide structure is transformed into a horizontally aligned rectangular waveguide structure. In the depicted embodiment, each of the core layers 4, 6 and 8 has a width w ₁ at the output end. At the output of structure 2, the width of the waveguide is w ₂ , which corresponds approximately to the sum of heights h ₁ , h ₂ and h ₃ . The height of the output is the height h ₂ of the middle layer. However, the geometry of the structure can vary depending on the requirements of the application.

A cladding having a lower refractive index than the core is usually arranged around the core to create light confinement.

Removing material from the upper core layer 4 and lower core layer 8 and adding material to the middle core layer 6 to achieve the transition, rotation of the waveguide axis can be simulated.

分量的灰阶图像。图3B表示开始结构16，中间结构18和终止结构20的基模电场分布中

分量的灰阶图像。3A-3B are grayscale images of the fundamental mode field propagating in the waveguide of the present invention. Figure 3A shows the fundamental mode electric field distribution of the starting structure 10, the intermediate structure 12 and the terminating structure 14

Component grayscale image. Fig. 3B shows the fundamental mode electric field distribution of the starting structure 16, the intermediate structure 18 and the terminating structure 20

Component grayscale image.

方向偏振，而中点是沿结构方向，模场分量是略微均匀分割，而在结构的终端，该模式主要是沿

方向偏振。耦合局部模理论指出，这些结构之间的绝热过渡能使最初波导中的

偏振态模式转换到最终波导2中的

偏振态。由于一次模式和二次模式都变换，反之亦然。即，初始波导中的

偏振态通常变换成最终波导中的

偏振态。此外，互易性原理确保器件可以相反运行。即，从结构终端开始的

偏振态和偏振态分别绝热变换成该结构首端的

偏振态和

偏振态。According to the mode distribution, it can be seen that the mode is initially along the

direction of polarization, while the midpoint is along the structure direction, the mode field components are split slightly evenly, while at the ends of the structure, the mode is mainly along

Directional polarization. Coupled localized mode theory states that the adiabatic transition between these structures enables the initial waveguide

Polarization mode conversion to the final waveguide 2 in the

polarization state. Since both primary and secondary modes are transformed, vice versa. That is, in the initial waveguide

The polarization state is usually transformed into the final waveguide

polarization state. Furthermore, the principle of reciprocity ensures that the devices can operate in reverse. That is, starting from the structure terminal

polarization state and The polarization states are adiabatically transformed into the

polarization state and

polarization state.

是线性的，然而，在其他的实施例中，这些参数可以不同，且可以利用非线性过渡。三维模式散射公式用于完成模拟，而所考虑的波长是1.55μm。在只要求每个波导横截面少量模式代表系统情况下，模式散射模拟是特别准确的建模工具。由于辐射模式基本不影响基于模式演变的方法进行，模式散射模拟非常适合于这些问题。此外，给出的结果得到确认，在几个器件长度下有完全的三维有限差分时域(FDTD)模拟。FDTD方法是完全数字实施Maxwell方程。图4中给出的模式散射模拟结果说明，等于或大于99％的功率成功地从TM偏振转移到TE偏振，其中变细长度仅仅是几百微米。假设波导是纯扭曲，我们发现，在变细长度太短时，因此，扰动太强而不能发生模式移相，导模交换功率和器件性能退化，使功率留在TM偏振态。FIG. 4 is a graph showing the performance of an embodiment of the converter of FIG. 2 as a function of structure length. In this example, h ₁ =h ₂ =h ₃ =w ₁ =0.25 μm, w ₂ =0.75 μm, the core refractive index is 2.2, and the cladding refractive index is 1.445, and the transition is along the direction of propagation

is linear, however, in other embodiments these parameters may be different and non-linear transitions may be utilized. The three-dimensional mode scattering formula was used to complete the simulation, and the considered wavelength was 1.55 μm. Mode scattering simulations are particularly accurate modeling tools where only a small number of modes per waveguide cross-section are required to represent the system. Model scattering simulations are well-suited for these problems since the radiation pattern is largely unaffected by the method based on pattern evolution. Furthermore, the presented results are confirmed with full three-dimensional finite-difference time-domain (FDTD) simulations at several device lengths. The FDTD method is a fully numerical implementation of Maxwell's equations. The mode scattering simulation results presented in Fig. 4 illustrate that equal or greater than 99% of the power is successfully transferred from TM to TE polarization with a taper length of only a few hundred micrometers. Assuming that the waveguide is a pure twist, we find that when the taper length is too short, and therefore, the perturbation is too strong for mode phase shifting, the guided mode exchanges power and degrades device performance, leaving power in the TM polarization state.

FIG. 5 is a graph showing broadband performance of the specific embodiment in FIG. 4 when the device length is set to 200 μm. Here, the 3D model scattering formulation is used to complete the simulation. Figure 5 illustrates that there is no discernible wavelength sensitivity over the entire range of 1.45µm to 1.65µm, which is a suitable wavelength range for telecommunications. The broadband performance of the method is consistent with the basic theory of operation. Since our goal is to transition from one mode state to another without introducing inter-mode coupling, the bandwidth is limited only by increasing the inter-mode coupling by one wavelength. At shorter wavelengths, the presence of additional modes can facilitate this coupling, while at longer wavelengths the individual modes become closely phase-matched, reducing the main effect of prohibiting coupling between modes. In any case, the For these phenomena to be effective, large changes in wavelength are required. This is in contrast to methods based on coupled modes, which suffer from inherent bandwidth limitations related to differences in supermode dispersion.

Many variations of the basic structure are possible. The geometry and refractive index may vary from the specific embodiments described above.

FIG. 6 is a schematic diagram of another embodiment of the polarization converter 24 of the present invention. Processing limitations may prevent the upper layer 26 and lower layer 28 from smoothly reaching infinitely small widths. Therefore, it is advantageous to consider a structure in which the final transition is separated from the middle layer 30 by means of the upper layer 26 and the lower layer 28 , as shown in FIG. 6 . This results in an approximately quite adiabatic transition to the final output waveguide.

The structure 24 is designed such that the upper 26, middle 30 and lower 28 initial heights are _hi , _h2 and _h3, respectively. In addition, each of the upper layer 26, the middle layer 30 and the lower layer 28 has a width _w1 at the input. Note that at the output of structure 24, the width is w ₂ , which roughly corresponds to the sum of heights h ₁ , h ₂ and h ₃ . The height at the output is the height h ₂ of the middle layer.

A cladding having a lower refractive index than the core is usually arranged around the core to create light confinement.

FIG. 7 is a schematic diagram of another embodiment of the polarization converter 32 of the present invention with properties similar to those in FIG. 6 . Here, however, the upper layer 34 and the lower layer 38 do not actually taper, but gradually separate from the middle layer 36 . In this way, the minimum feature size can be made larger, further facilitating processing. The upper layer 34 is separated from the middle layer 36 by a distance s at the output, and the middle layer 36 is separated from the lower layer 38 by a distance s at the output.

The structure 32 is designed such that the upper 34, middle 36 and lower 38 initial heights are _hi , _h2 and _h3, respectively. In addition, each of the upper layer 34, the middle layer 36 and the lower layer 38 has a width w ₁ . Note that at the output of structure 32, the width is w ₂ , which roughly corresponds to the sum of heights h ₁ , h ₂ and h ₃ . The height at the output is the height h ₂ of the middle layer.

A cladding having a lower refractive index than the core is usually arranged around the core to create light confinement.

FIG. 8 is a graph of the polarization converter performance of the embodiment of FIG. 7 as a function of structure length. In this embodiment, the parameters are set as follows: h ₁ =h ₂ =h ₃ =0.25 μm, w ₁ =0.25 μm, w ₂ =0.75 μm, s=0.125 μm and the refractive indices of the core and cladding layers, respectively are 2.2 and 1.445. In other embodiments, these parameters may be different. Again, the three-dimensional mode scattering formulation was used to complete the simulation, while the considered wavelength was 1.55 μm. With device lengths of only a few hundred microns, performance similar to the original embodiment can be obtained.

Fig. 9 is a graph showing broadband performance of the specific embodiment in Fig. 7 when the structure length is set to 100 μm. Here, the 3D model scattering formulation is also used to complete the simulation. Figure 9 illustrates that there is no discernible wavelength sensitivity over the entire range from 1.45 μm to 1.65 μm.

Although three layers is the minimum number required for a device to have symmetry, any of the above methods can also be applied to devices constructed of only two core layers. FIG. 10 is a schematic diagram of a polarization converter 40 utilizing only two core layers 42 and 44 . In this embodiment, the upper layer 42 and the lower layer 44 are respectively tapered and thickened and separated at the same time.

The polarization converter 40 is designed such that the initial heights of the upper layer 42 and the lower layer 44 are _hi and _h2 , respectively. In addition, each of the upper layer 42 and the lower layer 44 has a width w ₁ at the input. At the output, the upper layer 42 has a width w ₃ . Note that at the output end of the helical waveguide structure 40, the width is w ₂ , which roughly corresponds to the sum of the heights h ₁ and h ₂ . The height of the output is the height h ₂ of the lower layer. The performance of the device is not affected by the order of these two layers (ie, it does not matter which layer is on top).

A cladding having a lower refractive index than the core is usually arranged around the core to create light confinement.

Figure 11 is a graph of the performance of the structure of the embodiment of Figure 10 as a function of the length of the structure. In this embodiment, the parameters are set as follows: h ₁ =h ₂ =0.4 μm, w ₁ =0.4 μm, w ₂ =0.8 μm, w ₃ =0.25 μm, s=0.25 μm, and core and cladding The refractive indices are 2.2 and 1.445, respectively. In other embodiments, these parameters may be different. Again, the three-dimensional mode scattering formulation was used to complete the simulation, while the considered wavelength was 1.55 μm. Despite the inherent asymmetry of the two-layer embodiment, the structure does a fairly good job of transferring more than 99% of the power from TM polarization to TE polarization for structures only a few hundred micrometers in length.

FIG. 12 is a graph of broadband performance for the 100 μm long embodiment simulated in FIG. 11 . Likewise, the 3D model scattering formulation was used to complete the simulation. Figure 12 illustrates that there is no discernible wavelength sensitivity over the entire range from 1.45 μm to 1.65 μm.

Although the invention has been shown and described with reference to several preferred embodiments, various changes, omissions and additions may be made in form and detail thereof without departing from the scope of the invention.

Claims (12)

1. An integrated optical polarization converter having a length from an optical input section to an optical output section, It is characterized in that the polarization converter comprises: a plurality of core layers and a cladding layer, the cladding layer has a refractive index lower than that of the core layer and is arranged around the core layer; wherein the plurality of core layers are vertically aligned in the light output portion; wherein at least one of said core layers is laterally tapered along the length of the polarization converter, and/or is laterally separated from at least one other of said core layers; and The plurality of core layers is configured to simulate a gradually twisted waveguide, and wherein a propagating mode is adiabatically transformed from an initial polarization state to a different final polarization state. 2. The polarization converter of claim 1, wherein at least one of said core layers is laterally tapered linearly along the length of the polarization converter. 3. The polarization converter according to claim 1, wherein said plurality of core layers is composed of two core layers. 4. The polarization converter according to claim 1, wherein said plurality of core layers is no more than three core layers. 5. The polarization converter of claim 1, wherein a substantial number of said core layers have cross-sections that remain constant along the length of the polarization converter. 6. The polarization converter according to claim 1, wherein all of the core layers are tapered along the length of the polarization converter. 7. The polarization converter according to claim 1, wherein said plurality of core layers is composed of two core layers, both of which are tapered along the length of the polarization converter. 8. A polarization converter according to claim 7, wherein the tapering of one of the two core layers along the length of the polarization converter is opposite to the tapering of the other of the two core layers. 9. The polarization converter of claim 1, wherein said plurality of core layers are configured to geometrically simulate a gradually twisted waveguide. 10. A method of utilizing an integrated optical polarization converter, the method comprising: receiving an initial state of polarization at the input of the polarization converter; and forming a plurality of core layers and a cladding layer, the cladding layers having a refractive index lower than that of the core layer and arranged around the core layer, wherein the plurality of core layers are vertically aligned at the light output portion, and wherein at least one of the core layers layers are laterally tapered along the length of the polarization converter, and/or are laterally separated from at least one other of said core layers, and the plurality of core layers are configured to simulate a gradually twisted waveguide, and wherein from an initial The polarization state adiabatically transforms the propagating mode to a different final polarization state. 11. The method of claim 10, wherein said plurality of core layers is comprised of two core layers. 12. The method according to claim 10, wherein said plurality of core layers is no more than three core layers.

Images