Bandwidth equalization structure of electro-optical modulator and electro-optical modulator

Abstract

The invention provides an electro-optical modulator bandwidth equalization structure and an electro-optical modulator, and relates to the technical field of optical communication. The electro-optical modulator bandwidth balancing structure comprises a traveling wave electrode and N modulation sections distributed along the traveling wave electrode, the traveling wave electrode is divided into N continuous electrode sections in the optical signal propagation direction, N is an integer larger than 1, each electrode section corresponds to one modulation section, an active area load with a corresponding proportion is arranged in each modulation section, and the active area loads are distributed in the corresponding proportion. A segmented load structure is formed; wherein the active area load proportion in each modulation section is gradually reduced along the signal transmission direction, so that the high modulation efficiency is kept at the front end of signal propagation, the attenuation of the signal bandwidth is reduced at the rear end of signal propagation, and the electrode impedance is gradually increased. The scheme provided by the invention has the advantages that peaking improvement of high-frequency response is realized, and the overall bandwidth equalization effect is remarkable.

Description

Technical Field

[0001] This application relates to the field of optical communication technology, and more specifically, to an electro-optic modulator bandwidth equalization structure and an electro-optic modulator. Background Technology

[0002] Electro-optic modulators are core components in high-speed optical interconnects and coherent optical communication systems, and their performance directly determines system bandwidth, power consumption, and integration. With the rapid development of artificial intelligence (AI) technology, the demand for computing power is growing exponentially, driving a huge demand for AI computing clusters, especially GPU (Graphics Processing Unit) clusters. AI computing requires massively parallel GPU interconnects, placing higher demands on the single-wavelength rate of optical interconnects. Currently, the mainstream single-path rate of optical interconnects is 100Gbps, the industry is rapidly deploying 200Gbps single-path technology, while 400Gbps single-path has become a hot research topic and a key technological bottleneck for next-generation high-speed optical interconnects.

[0003] To achieve 400Gbps signal modulation in a single channel, it is essential to overcome the bandwidth bottleneck of electro-optic modulators, which strongly relies on high-bandwidth electro-optic modulator technology. Currently, most mainstream high-speed modulators employ a traveling-wave electrode (TWE) structure, propagating the radio frequency (RF) signal along the electrode in a quasi-traveling wave manner and engaging in electro-optic interactions with coplanar or clad-integrated optical waveguides (such as the Pockels effect based on LiNbO3, the plasmonic dispersion effect based on silicon or InP, or the quantum confinement Stark effect). To achieve high modulation efficiency and broadband response, the traveling-wave electrode must meet two fundamental conditions: optical-electrical phase velocity matching and impedance matching (typically targeting 50 Ω).

[0004] However, in practical devices, it is difficult to maintain the aforementioned ideal conditions simultaneously across the entire operating frequency band. Especially in integrated electro-optic modulators based on semiconductor materials (such as InP, SiGe, or SOI), the active region typically covers the entire modulation area in a continuous or uniform distribution. This structure results in a nearly constant capacitance per unit length of the traveling-wave electrode along the propagation direction, while transmission line losses (mainly composed of conductor ohmic losses and dielectric / radiation losses) increase with frequency, causing severe attenuation of high-frequency components at the rear of the electrode. Simultaneously, because the optical signal is continuously modulated, the rear optical field still needs to respond to the attenuated RF signal, resulting in a typical low-pass characteristic in the overall frequency response. The 3dB bandwidth is limited by high-frequency response collapse, rather than the modulation efficiency in the DC or low-frequency bands.

[0005] In summary, existing electro-optic modulator technologies cannot meet the requirements for signal integrity and bandwidth balance at single-channel speeds of 400Gbps and higher. Summary of the Invention

[0006] The purpose of this application is to provide an electro-optic modulator bandwidth equalization structure and an electro-optic modulator to solve the problem that existing electro-optic modulator technologies cannot meet the requirements for signal integrity and bandwidth equalization at single-channel speeds of 400Gbps and higher.

[0007] To achieve the above objectives, the technical solutions adopted in the embodiments of this application are as follows: On one hand, this application provides an electro-optic modulator bandwidth equalization structure, which includes traveling-wave electrodes and N modulation segments distributed along the traveling-wave electrodes. The traveling-wave electrodes are divided into N consecutive electrode segments along the optical signal propagation direction, where N is an integer greater than 1. Each electrode segment corresponds to a modulation segment, and each modulation segment is provided with a corresponding proportion of active region load, forming a segmented load structure; wherein, The proportion of active load in each modulation section gradually decreases along the signal transmission direction to maintain high modulation efficiency at the front end of signal propagation, reduce signal bandwidth attenuation at the back end of signal propagation, and gradually increase electrode impedance.

[0008] Optionally, the proportion of active region load in each modulation segment decreases gradually along the signal transmission direction.

[0009] Optionally, the difference in the active region load ratio between any two adjacent modulation sections is distributed in a gradient increasing manner along the signal transmission direction.

[0010] Optionally, along the signal propagation direction, the active region load ratio in the first part of the modulation section remains unchanged, while the active region load ratio in the second part of the modulation section gradually decreases.

[0011] Optionally, the N electrode segments are of equal length.

[0012] Optionally, the lengths of the N electrode segments gradually decrease along the signal transmission direction, and the ratio of the decrease in length of each electrode segment is equal to the ratio of the decrease in the active region load ratio within the corresponding modulation segment.

[0013] Optionally, the electro-optic modulator bandwidth equalization structure further includes a terminal load; When the terminal load is a matched load, the impedance of the terminal load is equal to the characteristic impedance of the traveling wave electrode to achieve optimal signal quality. When the terminal load is a mismatched load, the impedance of the terminal load is lower than the characteristic impedance of the traveling wave electrode, thereby enhancing the high-frequency response of the electro-optic modulator.

[0014] Optionally, the gradual reduction of the active region load ratio in each modulation section is achieved through distributed bias voltage; wherein, along the signal transmission direction, the reverse bias voltage applied to each modulation section gradually increases, so as to equivalently achieve the sequential reduction of the active region load ratio.

[0015] Optionally, the active region load is a silicon photonic modulator active region or a III-V compound semiconductor modulator active region.

[0016] On the other hand, embodiments of this application provide an electro-optic modulator, which includes the above-described electro-optic modulator bandwidth equalization structure.

[0017] Compared with the prior art, this application has the following advantages: This application provides an electro-optic modulator bandwidth equalization structure and an electro-optic modulator. The electro-optic modulator bandwidth equalization structure includes a traveling wave electrode and N modulation segments distributed along the traveling wave electrode. The traveling wave electrode is divided into N consecutive electrode segments along the optical signal propagation direction, where N is an integer greater than 1. Each electrode segment corresponds to a modulation segment. Each modulation segment is provided with a corresponding proportion of active area load, forming a segmented load structure. The proportion of active area load in each modulation segment gradually decreases along the signal transmission direction to maintain high modulation efficiency at the front end of signal propagation, reduce signal bandwidth attenuation at the back end of signal propagation, and gradually increase electrode impedance.

[0018] On the one hand, this application divides the traveling wave modulator into N modulation segments and finely controls the load ratio of each segment. Different numbers of segments N can be selected according to actual needs, achieving fine-grained bandwidth equalization adjustment, thus resulting in high bandwidth equalization accuracy and a high degree of design freedom. On the other hand, by adopting a gradually decreasing load distribution method, the impedance changes slowly along the signal transmission direction, effectively eliminating signal reflections introduced by impedance discontinuities and avoiding signal distortion problems caused by load mismatch in traditional load peaking techniques. This significantly improves signal quality, achieving excellent signal quality and good reflection suppression. Simultaneously, by reducing the load ratio at the back end, both signal bandwidth attenuation and gradual increase in electrode impedance are reduced, achieving peaking enhancement of the high-frequency response and a significant overall bandwidth equalization effect.

[0019] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram of the first structure of the electro-optic modulator bandwidth equalization structure provided in the embodiments of this application.

[0022] Figure 2 A schematic diagram of the bandwidth equalization response curve of the electro-optic modulator bandwidth equalization structure provided in the embodiments of this application.

[0023] Figure 3 This is a schematic diagram of a second structure of the electro-optic modulator bandwidth equalization structure provided in the embodiments of this application.

[0024] Figure 4 This is a schematic diagram of a third structure of the electro-optic modulator bandwidth equalization structure provided in the embodiments of this application.

[0025] Figure 5 This is a schematic diagram of a fourth structure of the electro-optic modulator bandwidth equalization structure provided in the embodiments of this application.

[0026] In the picture: 110 - Traveling wave electrode; 111 - Electrode segment; 120 - Modulation segment. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0028] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0029] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0030] It should be noted that in this paper, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0031] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0032] As described in the background section, the demand for computing power in AI is growing rapidly, almost doubling every three to four months. This has driven a huge demand for AI infrastructure, namely AI computing clusters. AI computing is large-scale parallel computing using GPU computing chips, requiring extensive and comprehensive interconnection, including interconnection of front-end networks, back-end scale-out networks, and back-end scale-up networks. Scale-up networks are interconnection networks between chips, especially between GPUs. Scale-up networks enable direct interconnection between GPU chips, allowing for pooling and resource sharing of computing chips (GPUs) and storage chips (especially HBM - High Bandwidth Memory). Expanding the scale of scale-up networks can significantly improve the computing efficiency of AI computing clusters. To better expand the scale of scale-up networks, it is necessary to significantly increase the single-wavelength rate and number of optical interconnects.

[0033] Currently, the mainstream single-path rate for optical interconnects is 100Gbps, and the industry is rapidly developing and deploying single-path 200Gbps technology. Current 100Gbps technologies include EML and silicon photonics, both of which do not face significant challenges in supporting single-path 200Gbps signal rates. However, with further increases in speed in the future, solving the signal modulation problem for single-path 400Gbps has become a hot technology in the industry. Achieving higher single-path rates strongly relies on high-bandwidth electro-optic modulator technology.

[0034] The mainstream methods for increasing the bandwidth of traveling wave modulators include: (1) adjusting the relationship between the characteristic impedance of the traveling wave electrode and the load resistance of the electro-optic modulator, generally using a method where the characteristic impedance of the electrode is greater than the load impedance to achieve the overall electro-optic bandwidth increase. (2) In addition to the above methods, the overall electro-optic bandwidth can also be increased by controlling the distribution of the bias voltage of the traveling wave electrode itself.

[0035] Both approaches have some drawbacks. For approach 1, the introduction of load bandwidth peaking technology can lead to signal distortion and degrade signal quality. Approach 2 results in a more complex modulator control circuit and poorer overall consistency, posing significant challenges for mass production and commercialization.

[0036] In summary, existing electro-optic modulator bandwidth equalization techniques suffer from several drawbacks, including signal quality degradation (sacrificing signal quality for bandwidth improvement), complex peripheral control circuitry hindering practical application, and poor consistency posing challenges to commercialization. Therefore, to address these issues, this application provides an electro-optic modulator bandwidth equalization structure based on modulator structure segmentation and single-segment distributed load control equalization techniques, effectively resolving the shortcomings of current modulator bandwidth equalization solutions. Furthermore, this structure offers advantages such as high implementability, no need for complex control, no signal quality degradation, and wide adjustability. The following is an illustrative description of the electro-optic modulator bandwidth equalization structure provided in this application: As an optional implementation, please refer to Figure 1 The bandwidth equalization structure of the electro-optic modulator includes a traveling wave electrode 110 and N modulation segments 120 distributed along the traveling wave electrode 110. The traveling wave electrode 110 is divided into N consecutive electrode segments 111 along the optical signal propagation direction, where N is an integer greater than 1. Each electrode segment 111 corresponds to a modulation segment 120. Each modulation segment 120 is provided with a corresponding proportion of active area load, forming a segmented load structure. The proportion of active area load in each modulation segment 120 gradually decreases along the signal transmission direction to maintain high modulation efficiency at the front end of signal propagation, reduce signal bandwidth attenuation at the back end of signal propagation, and gradually increase electrode impedance.

[0037] Understandably, electro-optic modulators play a crucial role in high-speed optical interconnect systems by efficiently converting electrical signals into optical signals. Their performance directly limits whether the single-wavelength rate can be increased to 200Gbps or even 400Gbps. However, the uniform structure traveling-wave electrode 110 commonly used in existing technologies suffers from a significant decrease in the actual usable bandwidth in the later region during signal propagation due to the continuous attenuation of the radio frequency signal on the electrode. At the same time, signal reflection is caused by abrupt changes in load impedance between the front and rear sections, which degrades the quality of the modulated signal. This problem essentially stems from the fact that traditional designs treat the entire traveling-wave electrode 110 as a single, homogeneous loading unit, failing to adapt it to the dynamic amplitude and bandwidth characteristics of the signal along the propagation path.

[0038] Therefore, this application divides the traveling wave electrode 110 into N consecutive electrode segments 111 along the optical signal propagation direction, and assigns a modulation segment 120 to each electrode segment 111. Each modulation segment 120 contains a corresponding proportion of active region loads, thus forming a segmented load structure. N is an integer greater than 1, and its value is determined based on the fineness of bandwidth equalization: when precise adjustment is required, N can be a larger value, such as 15; when only coarse-grained adjustment is required, N can be a smaller value, such as 5; in a typical embodiment, N is 10. In one implementation, the N electrode segments 111 have equal lengths. For example, if the overall length of the modulator is 3 mm, then the length of each electrode segment 111 is 0.3 mm. In another implementation, the lengths of the N electrode segments 111 can also be unequal and gradually decrease along the signal transmission direction. For example, the length of the first electrode segment 111 is 0.3 mm, the length of the second electrode segment 111 is 0.29 mm, and so on.

[0039] It should be noted that in this application, the active region load ratio in each modulation segment 120 gradually decreases along the signal transmission direction. At the front end of signal propagation, i.e., the first electrode segment 111 and its corresponding modulation segment 120, since the input signal has not yet undergone significant attenuation and the bandwidth is high, a higher active region load ratio is configured to fully utilize the high bandwidth to achieve efficient electro-optic modulation. As the signal passes through the second, third... up to the Nth electrode segment 111 from front to back along the propagation direction, the signal amplitude and effective bandwidth gradually attenuate. At this time, the active region load ratio in each subsequent modulation segment 120 is gradually reduced. On the one hand, this can weaken the signal attenuation trend on the traveling wave electrode 110, thereby improving the response capability of the later region to high-frequency components. On the other hand, it can make the equivalent impedance of the electrode gradually increase along the transmission direction, forming a smooth and continuous impedance gradient, avoiding signal reflection caused by impedance abrupt changes, and thus balancing overall bandwidth improvement and signal quality maintenance.

[0040] In this process, the active region load refers to the active region or modulation region used to realize the electro-optic modulation function. The change of its load ratio can be achieved by physically changing the actual coverage area of ​​the active region within each modulation segment 120, or by adjusting the equivalent load through distributed bias voltage. Along the signal transmission direction, the reverse bias voltage applied to each modulation segment 120 gradually increases to equivalently reduce the active region load ratio sequentially. Regardless of the implementation method, this structure does not rely on the mismatch of the terminal load, and therefore does not need to sacrifice the matching condition to obtain bandwidth peaking. This fundamentally avoids the problems of increased return loss and eye diagram closure caused by terminal impedance mismatch in traditional peaking technology.

[0041] The specific bandwidth equalization effect of the electro-optic modulator bandwidth equalization structure provided in this application is as follows: Figure 2 As shown, this represents a typical equalization bandwidth, with frequency on the horizontal axis and response amplitude on the vertical axis. The response amplitude exhibits a peak (peaking) in the mid-to-high frequency range, meaning the amplitude is higher than adjacent frequency bands at a certain frequency, forming a local maximum. After peaking, the response amplitude gradually decreases at higher frequencies. Furthermore, the more segments N there are, the higher the bandwidth equalization resolution of the modulator. The load of each segment is reduced more than the load of the previous segment, and the higher the frequency range is raised. Simultaneously, the negative effects of reflections are reduced.

[0042] Combination Figure 2 Yes, in this application, by segmenting the traveling wave electrode 110 and setting a gradually decreasing active region load ratio, a bandwidth peaking mechanism is achieved. Specifically, the reduced load in the later stage reduces the transmission loss of high-frequency signals and compensates for the problem of rapid attenuation of high-frequency bands in traditional modulators; the impedance gradient structure forms a distributed matching network, enabling high-frequency components to obtain additional transmission gain; and the multi-segment cascaded distributed filtering effect achieves controllable bandwidth peaking, effectively extending the 3dB bandwidth and optimizing the in-band flatness.

[0043] Bandwidth peaking offers the following advantages: 1. Effectively Extended 3dB Bandwidth: Traditional traveling-wave modulators exhibit typical low-pass characteristics in their frequency response, with rapid amplitude decay in the high-frequency band, limiting the 3dB bandwidth due to high-frequency response collapse. This application introduces bandwidth peaking to create local gain in the high-frequency band, compensating for high-frequency attenuation and significantly extending the 3dB bandwidth. This is crucial for achieving single-channel optical interconnects with speeds of 400Gbps and higher.

[0044] 2. Achieving In-Band Flatness Optimization: The core purpose of peaking is not to pursue infinitely high peak values, but to make the response amplitude variation more gradual throughout the entire operating frequency band through a combination of high-efficiency modulation in the low-frequency band and peaking compensation in the high-frequency band. Improved in-band flatness means reduced signal distortion at different frequency components, which is beneficial for high-quality transmission of high-speed signals.

[0045] 3. Avoiding the side effects of traditional peaking techniques: Traditional load peaking techniques generate peaking effects by introducing signal reflection through terminal load mismatch (such as electrode characteristic impedance being greater than load impedance), but reflection leads to signal distortion and quality degradation. This application achieves peaking through a segmented gradient load structure, which does not rely on terminal reflection, thus maintaining excellent signal quality while achieving bandwidth expansion.

[0046] Therefore, the segmented load structure provided in this application does not simply divide the electrodes into multiple segments, but rather constructs a collaborative response mechanism based on the dynamic attenuation law of the signal along the propagation path, which ensures efficiency with high load at the front end and suppresses attenuation with low load at the back end. It not only solves the problem of uneven overall response caused by non-uniform attenuation of signal bandwidth along the propagation direction, but also overcomes the reflection distortion problem caused by impedance discontinuity in traditional load peaking technology. Thus, it achieves adjustable and balanced bandwidth characteristics of the electro-optic modulator without increasing the complexity of external control or degrading signal quality.

[0047] In terms of implementation, this application can adopt various methods to achieve a gradual decrease in the active area load ratio in each modulation section 120 along the signal transmission direction.

[0048] In the first implementation, this embodiment configures different proportions of active region loads for N modulation segments 120 distributed along the signal transmission direction in the electro-optic modulator bandwidth equalization structure, and makes the proportion of the active region loads decrease in a gradient manner, thereby synergistically improving the technical problem of high-frequency response attenuation and overall bandwidth flatness without significantly increasing reflection or destroying signal integrity.

[0049] like Figure 1 As shown, the gradient decreasing distribution described in this application refers to the active area load ratio within each modulation segment 120 decreasing arithmetically or proportionally along the signal transmission direction. That is, the load ratio difference between any two adjacent modulation segments 120 is a constant value, or the load ratio between any two adjacent modulation segments 120 is a constant value. This distribution method has a clear mathematical regularity, facilitating design and manufacturing implementation. During design, the active area load ratio of the first segment and the active area load ratio of the last segment are first determined. Then, based on the number of segments N, the decreasing amount of each segment is calculated, so that the load ratios of each segment form an arithmetic or geometric sequence.

[0050] Taking N=10 as an example, assuming the first load segment has a load ratio of 100% and the last load segment has a load ratio of 10%, using an arithmetic progression method: total reduction = 100% - 10% = 90%, reduction per segment = 90% ÷ 9 = 10%. The load ratios for each segment are: 100%, 90%, 80%, 70%, 60%, 50%, 40%, 30%, 20%, 10%. If a geometric progression method is used, the load ratio for each segment is a fixed ratio of the previous segment (e.g., 0.8 times), and the load ratios for each segment are: 100%, 80%, 64%, 51.2%, 41.0%, 32.8%, 26.2%, 21.0%, 16.8%, 13.4%.

[0051] By adopting a gradient-decreasing distribution, the design offers several advantages. First, it provides strong predictability, as the gradient-decreasing distribution follows clear mathematical rules and requires fewer design parameters. Only the initial load ratio, the final load ratio, and the decreasing pattern need to be determined, avoiding complex step-by-step optimization. Second, its impedance change is smooth. The arithmetic or geometrical decrease ensures a smooth change in the characteristic impedance of the traveling wave electrode 110 along the transmission direction, reducing impedance discontinuities, minimizing signal reflection, and maintaining signal quality. Furthermore, its bandwidth equalization effect is stable. The regular decreasing distribution provides stable frequency response characteristics for the distributed filtering effect, and the position and amplitude of bandwidth peaking are well-controllable, facilitating collaborative design with other optoelectronic devices.

[0052] For the second implementation, please refer to Figure 3 The difference in the active region load ratio within each two adjacent modulation sections (120) increases gradually along the signal transmission direction. In other words, the decrease in load ratio gradually increases along the signal transmission direction. The core of this distribution is that at the signal transmission front end, the signal strength is high, and the load ratio decreases less to ensure modulation efficiency; at the signal transmission back end, signal attenuation intensifies, and the load ratio decreases more significantly to further reduce back-end losses and increase impedance, thereby compensating for high-frequency attenuation.

[0053] Taking N=10 as an example, let the first segment's load ratio be 100% and the last segment's load ratio be 10%. Using an incremental difference method, let the differences between adjacent segments be 2%, 4%, 6%, 8%, 10%, 12%, 14%, 16%, and 18% respectively (totaling 90%). The load ratios of each segment are 100%, 98%, 94%, 88%, 80%, 70%, 58%, 44%, 28%, and 10% respectively. It can be seen that the load ratio decreases slowly in the first segment and rapidly in the second segment, forming a decreasing curve that is "gradual at the beginning and steep at the end".

[0054] By setting the difference in the active region load ratio within each adjacent modulation segment (120) to a gradient increasing distribution along the signal transmission direction, the following advantages are achieved: First, front-end modulation efficiency is maximized. The slow decrease in the front-end load ratio means that the signal transmission front-end maintains a high load ratio, thereby achieving efficient electro-optic modulation in the region of highest signal strength and maximizing overall modulation efficiency. Second, back-end high-frequency compensation is enhanced. The rapid decrease in the back-end load ratio significantly reduces the absorption loss of the RF signal in the back-end and more significantly increases the impedance of the back-end electrodes, resulting in a stronger high-frequency peaking effect and effectively compensating for high-frequency attenuation. Furthermore, this method can adapt to different frequency band compensation requirements. For applications with particularly severe high-frequency attenuation, the incremental difference distribution provides stronger back-end compensation capabilities and achieves a wider operating bandwidth.

[0055] For the third implementation, please refer to Figure 4 Along the signal propagation direction, the active region load ratio remains constant in the first modulation section 120, while the active region load ratio gradually decreases in the second modulation section 120. In other words, this implementation divides the modulator into two functional regions: a constant load region in the front and a decreasing load region in the back. The load ratio of each segment in the constant load region remains unchanged, while the load ratio of each segment in the decreasing load region gradually decreases along the signal transmission direction. Therefore, in the front-end region where the signal strength is high, a constant high load is used to maximize modulation efficiency; in the back-end region where signal attenuation is significant, a decreasing load is used to compensate for high-frequency attenuation and increase impedance.

[0056] Taking N=10 as an example, let the first 5 segments be the constant load zone and the last 5 segments be the decreasing load zone. In the first 5 segments, the load ratio is 100%. In the last 5 segments, starting from the 6th segment, the load decreases, let's say the 6th segment is 80%, and then 60%, 40%, 20%, and 10%. At this time, the overall load ratio distribution is: 100%, 100%, 100%, 100%, 100%, 80%, 60%, 40%, 20%, and 10%. Of course, the number of segments in the constant load zone and the decreasing load zone can be adjusted according to actual needs. For example, different combinations can be used, such as the first 3 segments being constant and the last 7 segments decreasing, or the first 7 segments being constant and the last 3 segments decreasing.

[0057] This decreasing method has at least the following effects: 1. Simplified Design Complexity: Compared to a full-section gradient reduction approach, the constant front-end load followed by a decreasing load reduces the number of parameters requiring precise control, making the design process simpler and facilitating rapid implementation and engineering applications. 2. Maximized Front-End Modulation Efficiency: Maintaining the highest load ratio in the front-end ensures the most efficient electro-optic modulation in the region of highest signal strength, avoiding sacrificing overall modulation efficiency due to reduced front-end load. 3. Balanced Back-End Compensation Needs: The decreasing load in the back-end achieves high-frequency compensation, forming a functional partition with the constant high load of the front-end. This ensures both modulation efficiency and bandwidth balance, achieving the dual benefits of improved efficiency and bandwidth equilibrium. 4. Good Process Tolerance: The constant front-end load ratio reduces sensitivity to process deviations; the decreasing load in the back-end also allows for more lenient process tolerance requirements, resulting in good mass production consistency for the overall solution.

[0058] For the fourth implementation, please refer to Figure 5 The lengths of the N electrode segments 111 gradually decrease along the signal transmission direction, and the ratio of the decrease in length of each electrode segment 111 is equal to the ratio of the decrease in the active region load ratio within the corresponding modulation segment 120.

[0059] Specifically, let the length of the i-th electrode segment 111 be L. i The load ratio of the active region in the i-th segment is R. i Then the following condition is met: (L) i - L i+1 ) / L i =(R i - R i+1 ) / R i ; or L i+1 / L i =R i+1 / R i ; Taking N=5 as an example, let the load ratios be 100%, 80%, 64%, 51.2%, and 41.0% respectively (decreasing proportionally with a common ratio of 0.8). Let the length of the first segment be 1.0mm. Then the lengths of each segment are 1.000mm, 0.800mm, 0.640mm, 0.512mm, and 0.410mm respectively.

[0060] This decreasing method has at least the following effects: 1. Further optimization of impedance variation: In the segmented load structure, the length of electrode segment 111 decreases in tandem with the load ratio, resulting in a more uniform rate of change of capacitance per unit length. Specifically, the capacitance per unit length of the traveling wave electrode 110 is mainly affected by the load ratio in the active region. When the length of electrode segment 111 changes proportionally with the load ratio, the contribution of each segment to the overall impedance variation remains consistent. This makes the impedance variation along the signal transmission direction smoother and avoids local impedance abrupt changes caused by the mismatch between segment length and load ratio. Compared with traditional equal-length segments, the tandemly decreasing scheme can further reduce impedance discontinuities and signal reflections.

[0061] 2. Enhancing the Synergy of Distributed Filtering Effects: The strength of distributed filtering effects is closely related to the electrical length of each segment (i.e., the product of physical length and propagation constant). When the length of electrode segment 111 decreases in tandem with the load ratio, the front segment exhibits a longer length and a higher load ratio, providing a stronger modulation effect; the rear segment exhibits a shorter length and a lower load ratio, providing weaker losses and a stronger impedance rise. This distribution of longer front and shorter rear, higher front and lower rear, makes the roles of each segment in the distributed filtering network clearer. The front segment, as the modulation segment, dominates the low-frequency response, while the rear segment, as the compensation segment, dominates the high-frequency peaking. The contributions of each segment complement each other in the frequency domain, thereby achieving a better bandwidth equalization effect.

[0062] 3. Improved Design Consistency and Scalability: The linked decreasing scheme establishes an analytical relationship between the length of electrode segment 111 and the load ratio. The design parameters are simplified from N independent length values ​​and N independent load ratio values ​​to N load ratio values ​​plus a baseline length value. This design method only requires determining the load ratio distribution, and the length of electrode segment 111 is automatically determined, thereby improving design efficiency. Furthermore, when the number of segments N needs to be adjusted, it is only necessary to recalculate according to the same proportional rule, without re-optimizing all parameters, resulting in strong parameter scalability. In addition, when the overall length of the modulator needs to be adjusted due to process requirements, the bandwidth balance characteristics can be maintained by scaling the length of each segment proportionally.

[0063] 4. Improved High-Frequency Response Uniformity: In traditional equal-length segmented designs, the length of the rear electrode segment 111 is the same as the front segment, but the RF signal wavelength of the rear segment may change due to factors such as dielectric loss. In the linked decreasing design, the length of the rear electrode segment 111 is shortened accordingly, ensuring that the electrical length of each segment remains relatively consistent in the frequency domain, avoiding electrical length mismatch problems caused by fixed physical lengths. This helps to stabilize the position of bandwidth peaking, reduce frequency response fluctuations caused by process variations, and improve product consistency.

[0064] 5. Optimize chip area utilization efficiency: For integrated optoelectronic chips with limited overall length, the linked decreasing scheme allocates the shorter electrode segment 111 to the later section. Since the later section itself has a lower load ratio and reduced modulation efficiency, it is less sensitive to length. Therefore, this scheme can effectively shorten the total length of the modulator without sacrificing overall performance, thereby improving chip area utilization efficiency and reducing manufacturing costs.

[0065] In summary, during practical applications, due to the inherent attenuation characteristics of optical signals propagating on the traveling-wave electrode 110, the signal does not attenuate significantly at the front end, resulting in a high bandwidth. Therefore, if a high proportion of active region load is applied to the front-end modulation section 120, efficient electro-optic modulation can be achieved by fully utilizing the high bandwidth. However, as the signal propagates further, both amplitude and bandwidth gradually decrease. Maintaining the same intensity of active region load at the front end would exacerbate signal attenuation on the traveling-wave electrode 110 and lead to further loss of high-frequency components. Therefore, by distributing the proportion of active region load in each modulation section 120 in a gradient decreasing direction along the signal transmission path, the loading effect of the active region on the traveling-wave electrical signal at the rear end can be reduced, decreasing the bandwidth attenuation rate of the signal in the later part of the propagation path. On the other hand, the reduction in the active region load ratio is equivalent to reducing the local capacitive load, thereby gradually increasing the characteristic impedance of the corresponding electrode segment 111, which in turn produces a peaking effect on the high-frequency response and compensates for the low-pass attenuation characteristics. In addition, this gradient reduction method is different from the traditional uniform load structure, avoiding strong signal reflection caused by sudden load changes, thus taking into account both bandwidth expansion and signal quality optimization. For example, when N takes a larger value (such as N≥8), the load change is more gradual, the bandwidth balance resolution is higher, the high-frequency response rise is more continuous, and the reflection interference is weaker. When the difference in the active region load ratio between every two adjacent modulation segments 120 increases in a gradient, it means that the load reduction in the later stage is more drastic, which is conducive to more significantly improving the high-frequency response amplitude.

[0066] Therefore, it can be seen that the active area load ratio in each modulation segment 120 is distributed in a gradient decreasing manner along the signal transmission direction, and the difference in the active area load ratio in each two adjacent modulation segments 120 is distributed in a gradient increasing manner along the signal transmission direction. The active area load ratio in the first part of the modulation segment 120 along the signal propagation direction remains unchanged, while the active area load ratio in the second part of the modulation segment 120 gradually decreases. In addition, the lengths of the N electrode segments 111 are equal. These features together constitute a systematic load configuration strategy that takes into account modulation efficiency, bandwidth balance and signal integrity. Its core lies in implementing a spatially non-uniform but regularly controllable gradient distribution of the active area load based on the dynamically changing bandwidth and amplitude characteristics of the signal during propagation. This ensures high modulation efficiency under high signal-to-noise ratio at the front end and alleviates the high-frequency attenuation and impedance mismatch problems caused by excessive load at the back end, ultimately achieving the broadening and flattening of the overall frequency response curve of the electro-optic modulator.

[0067] Furthermore, by gradually reducing the length of the N electrode segments 111 along the signal transmission direction, and by reducing the length of each electrode segment 111 by a ratio equal to reducing the ratio of the active region load ratio within the corresponding modulation segment 120, the bandwidth attenuation characteristics and load response requirements of the signal dynamically changing during propagation are synchronously matched in the spatial distribution dimension of the traveling wave electrodes 110, thereby achieving more refined bandwidth equalization control.

[0068] In practical applications, as the optical signal propagates on the traveling wave electrode 110, its high-frequency components attenuate rapidly with increasing distance. Simultaneously, the equivalent capacitance per unit length of the electrode and the load in the active region jointly determine the local impedance and phase velocity matching degree. If only the load ratio in the active region is adjusted while keeping the length of electrode segment 111 constant, although reducing the load in the later section can raise the impedance, the risk of phase mismatch and reflection still exists because the capacitance per unit length is not reduced accordingly. However, by synchronously reducing the length of electrode segment 111, on the one hand, the total parasitic capacitance of the later section electrode can be reduced, further alleviating high-frequency attenuation; on the other hand, the reduction in length and the reduction in load ratio are at the same ratio, ensuring that each modulation segment 12... The geometric-electrical dual coordination of the electro-optical interaction strength makes the modulation contribution of each segment to the signal more compatible with its remaining bandwidth capacity. Specifically, the front electrode segment 111 is longer and has a higher load ratio, which is conducive to achieving full modulation when the signal energy is abundant and the bandwidth is sufficient. The rear electrode segment 111 is shorter and has a lower load ratio. When the signal has attenuated and high-frequency components are scarce, it avoids the additional loss caused by overloading and suppresses slow wave effects and dispersion accumulation by shortening the physical length, thereby improving the flatness and extension of the overall frequency response. For example, when N is 8 or more, this gradient configuration of length and load can significantly improve the response consistency of the frequency band above 30 GHz, and compared with the structure of the electrode segment 111 of equal length, the reflection coefficient is reduced by about 40% and the bandwidth can be broadened by more than 25%.

[0069] It can be seen that the length of the N electrode segments 111 gradually decreases along the signal transmission direction, and the ratio of the length reduction of each electrode segment 111 is equal to the ratio of the reduction of the active area load ratio within the corresponding modulation segment 120. This feature is not an isolated geometric adjustment, but a systematic design that quantitatively couples the physical scale of the electrodes with the electrical load of the active area. Its technical effect is that it not only enhances the modulation depth under high signal-to-noise ratio conditions at the front end, but also optimizes the high-frequency fidelity capability under low amplitude conditions at the back end. Furthermore, by changing the length proportionally to the load, it naturally suppresses impedance discontinuities and signal reflections caused by parameter mutations. Ultimately, it achieves broadband, low-distortion, and high-efficiency integrated electro-optic modulation without adding additional matching components.

[0070] It should be noted that when the active region load ratio in each modulation segment 120 is gradually reduced through distributed bias voltage, in the bandwidth equalization structure of the electro-optic modulator, the active region size or density of each modulation segment 120 is not fixed by physical means such as etching, doping or geometric reconstruction. Instead, an independent and controllable DC bias circuit is connected to each modulation segment 120 corresponding to each traveling wave electrode 110, and a reverse bias voltage (i.e., reverse bias voltage) of different amplitude is applied to each modulation segment 120. The reverse bias voltage applied to each modulation segment 120 gradually increases along the signal transmission direction. In this process, "distributed" means that the bias voltage of each modulation segment 120 is provided by multiple bias nodes that are isolated from each other and can be adjusted independently, rather than setting a single bias point at the input or output end. Reverse bias voltage refers to the voltage polarity applied across a semiconductor pn junction or pin structure to broaden the depletion layer. Increasing this voltage leads to a decrease in carrier concentration and a reduction in optical absorption and refractive index modulation efficiency within the active region, thereby effectively reducing the electro-optic modulation contribution of this segment to the optical signal. In other words, it effectively achieves a sequential decrease in the load ratio of the active region. For example, in an embodiment where N=6, the first modulation segment 120 is subjected to… A 1.0 V reverse bias voltage is applied to the second modulation segment 120. 1.2 V, third modulation segment 120 applied 1.5 V, fourth modulation segment 120 applied 1.9 V, fifth modulation segment 120 applied 2.4 V, sixth modulation segment 120 applied 3.0 V, as the absolute value of the reverse bias voltage increases step by step, the equivalent active region load ratio of the corresponding modulation section 120 shows a continuous decreasing trend.

[0071] The active region load is either the active region of a silicon photonic modulator or the active region of a III-V compound semiconductor modulator. The active region of a silicon photonic modulator is usually based on a silicon-based pin or MOS structure, and phase modulation is achieved by utilizing the carrier depletion effect. The active region of a III-V compound semiconductor modulator is based on materials such as gallium arsenide and indium phosphide, and high-speed modulation can be achieved by relying on quantum wells or electro-absorption effects. Although the materials and physical mechanisms are different, both can control the depletion width and carrier distribution by reverse bias voltage, thereby affecting their equivalent load strength. Therefore, both are suitable for the distributed bias control method described in this scheme.

[0072] In practical applications, the load ratio of the active region of the electro-optic modulator directly affects the equivalent impedance, signal attenuation rate, and phase velocity matching state of the traveling wave electrode 110. If a fixed structural design is adopted (such as etching active regions of different lengths), it cannot be adjusted once the process is completed, making it difficult to meet different operating conditions or system requirements. However, by using distributed bias voltage to achieve a gradual change in the load ratio of the active region, the response characteristics of each segment can be flexibly adjusted after the device is packaged. For example, a smaller reverse bias gradient can be used in low-speed, high-fidelity scenarios to maintain overall linearity, while a larger reverse bias gradient can be used in high-speed, broadband scenarios to enhance the high-frequency boost effect. In addition, this method avoids interface reflection and stress concentration caused by abrupt changes in physical structure, improving manufacturing yield and long-term reliability. Specifically, the front-end modulation section 120 applies a lower reverse bias voltage to maintain a larger carrier modulation dynamic range, adapting to the initial signal state with high signal-to-noise ratio and high bandwidth; the back-end modulation section 120 applies a higher reverse bias voltage to compress the carrier response window and reduce the equivalent capacitance, which not only alleviates high-frequency attenuation but also synergistically raises the local electrode impedance, forming a smooth impedance transition and suppressing standing wave formation. For example, when using the active region of a silicon photonic modulator, the reverse bias voltage from... Gradually increase from 0.8 V to At 2.5 V, the equivalent active region load ratio can be linearly reduced from about 95% to about 40%. When using the active region of a III-V compound semiconductor modulator, the load ratio change curve is slightly different under the same reverse bias gradient due to the difference in the built-in potential and carrier mobility of the material, but it still follows the same control logic.

[0073] As can be seen, this application, on the one hand, replaces structural modifications with electrical means, significantly improving design flexibility and process compatibility; on the other hand, through the decoupling and adaptation of the reverse bias voltage gradient and the active region material characteristics, the same balanced architecture can be seamlessly migrated to silicon-based optoelectronic platforms or high-speed III-V integrated platforms, expanding application scenarios and industrialization value.

[0074] Furthermore, in one implementation, the electro-optic modulator bandwidth equalization structure also includes a terminal load; when the terminal load is a matched load, the impedance of the terminal load is equal to the characteristic impedance of the traveling wave electrode 110 to achieve optimal signal quality; when the terminal load is a mismatched load, the impedance of the terminal load is lower than the characteristic impedance of the traveling wave electrode 110 to enhance the high-frequency response of the electro-optic modulator.

[0075] By setting the terminal load in the bandwidth equalization structure of the electro-optic modulator and selectively setting between matched load and mismatched load configurations according to actual performance requirements, the dual goals of optimizing signal quality and improving high-frequency response can be achieved.

[0076] In practical applications, a matched load refers to a load whose impedance value is equal to the characteristic impedance of the traveling wave electrode 110 under normal operating conditions (i.e., the inherent impedance of the radio frequency waveguide under non-reflection conditions). At this time, the signal is almost completely absorbed after reaching the terminal, and the reflection approaches zero, thereby maximizing the integrity and fidelity of the modulated signal. A mismatched load, on the other hand, refers to a load whose impedance value is lower than the characteristic impedance of the traveling wave electrode 110. At this time, some signal energy will undergo controllable negative reflection at the terminal. The reflected wave is superimposed on the original signal under a specific phase relationship, which can effectively enhance the response amplitude of high-frequency components in the frequency domain, thereby widening the overall electro-optic bandwidth.

[0077] In this process, the two terminal load configurations mentioned above are not mutually exclusive, but can be optionally used in conjunction with the segmented load structure. In the specific embodiments described above, they can also be combined with traditional load balancing techniques to further improve the high-frequency response by using a lower modulator terminal load; alternatively, they can be used without combination, employing a matched terminal load to optimize the quality of the modulated signal. Thus, this application does not simply regard the terminal load as a single functional component, but rather as one of the programmable control dimensions, forming an orthogonal adjustment mechanism with the segmented active region load structure. The former controls the global reflection behavior and frequency response trend, while the latter controls the local impedance gradient and bandwidth equalization accuracy along the transmission path. Together, they avoid the severe signal distortion caused by relying solely on terminal mismatch and overcome the high-frequency response bottleneck caused by relying solely on matched terminals.

[0078] Based on the above implementation, this application embodiment also provides an electro-optic modulator, which includes the above-described electro-optic modulator bandwidth equalization structure.

[0079] In summary, the electro-optic modulator bandwidth equalization structure provided in this application includes at least the following technical features: (1) Divide the modulation region of the traveling wave electrode of the modulator into N equal segments. If fine control is required, N should be a larger value; if coarse control is required, N should be a smaller value. In practical applications, N segments can also be divided into segments of unequal length.

[0080] (2) In the N-segment segmentation, a gradually decreasing modulation area load scheme is adopted. This allows the front end of the modulator to adopt a high-load structure, which can improve the modulation efficiency. The bandwidth of the rear segment is reduced, and a low-load scheme is adopted, which can improve the response of the high-frequency part and increase the impedance value of the rear segment, thereby achieving high-frequency rise and bandwidth balance.

[0081] (3) After the entire modulation region is segmented, a gradient-based load reduction scheme is adopted, which allows the impedance rise to change slowly, greatly eliminating signal reflections caused by impedance discontinuities and thus improving signal quality. At the same time, the load reduction amplitude of each stage can be adjusted according to actual needs.

[0082] (4) This application can be applied to silicon photonic traveling wave modulators, as well as to three- to five-group traveling wave modulators. It is also compatible with existing terminal load bandwidth balancing technologies.

[0083] The electro-optic modulator bandwidth equalization structure provided in this application has at least the following technical effects: (1) High bandwidth equalization accuracy and high design freedom: By dividing the traveling wave modulator into N segments and finely controlling the load ratio of each segment, different numbers of segments N can be selected according to actual needs, realizing bandwidth equalization adjustment from coarse to fine. The N segments can be divided into equal or unequal lengths, and the load drop of each stage can also be flexibly set, thus providing rich design freedom and adapting to the bandwidth equalization needs of different application scenarios.

[0084] (2) Excellent signal quality and good reflection suppression: By gradually adjusting the impedance in stages, the rise in impedance can be made to change slowly, which can eliminate the signal reflection caused by impedance discontinuity to the greatest extent and avoid the signal distortion problem caused by load mismatch in traditional load peaking technology, thereby significantly improving signal quality. At the same time, the gradual decrease in load can further increase the response frequency of the high-frequency part, and achieve the maximum benefit of bandwidth balance.

[0085] (3) Simple control circuit and easy commercialization: Compared with the traditional distributed reverse bias voltage scheme, the load segmentation ratio adjustable scheme of the present invention has greater flexibility and a simpler implementation path. This scheme does not require a complex external bias distribution circuit, which can greatly simplify the chip's control circuit, is compatible with existing modulator control circuits and schemes, and has good prospects for mass production and commercialization. Even if a distributed bias voltage is used as an alternative implementation method, the bias change law of this scheme is based on the static preset value equivalent to the structural segmented load scheme, without the need for dynamic feedback control, and the control circuit is also greatly simplified.

[0086] (4) Strong compatibility and ability to achieve multiple bandwidth balancing: This solution achieves bandwidth balancing through segmentation of the modulator's modulation region and load adjustment, eliminating the need to rely on mismatched terminal load resistors to increase bandwidth and thus avoiding signal degradation caused by load reflection. Furthermore, this solution does not conflict with traditional terminal load bandwidth balancing mechanisms; the two can be used together to achieve stronger bandwidth balancing capabilities and meet the needs of applications with higher bandwidth requirements.

[0087] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0088] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from the spirit or essential characteristics of this application. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of this application is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within this application. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A bandwidth equalization structure for an electro-optic modulator, characterized in that, The electro-optic modulator bandwidth equalization structure includes traveling-wave electrodes and N modulation segments distributed along the traveling-wave electrodes. The traveling-wave electrodes are divided into N consecutive electrode segments along the optical signal propagation direction, where N is an integer greater than 1. Each electrode segment corresponds to one modulation segment, and each modulation segment contains a corresponding proportion of active region load, forming a segmented load structure. The proportion of active load in each modulation section gradually decreases along the signal transmission direction to maintain high modulation efficiency at the front end of signal propagation, reduce signal bandwidth attenuation at the back end of signal propagation, and gradually increase electrode impedance.

2. The electro-optic modulator bandwidth equalization structure according to claim 1, characterized in that, The proportion of active load in each modulation segment decreases gradually along the signal transmission direction.

3. The electro-optic modulator bandwidth equalization structure according to claim 1, characterized in that, The difference in the active region load ratio between any two adjacent modulation sections increases gradually along the signal transmission direction.

4. The electro-optic modulator bandwidth equalization structure according to claim 1, characterized in that, Along the signal propagation direction, the active region load ratio remains unchanged in the first part of the modulation section, while the active region load ratio gradually decreases in the second part of the modulation section.

5. The electro-optic modulator bandwidth equalization structure according to claim 1, characterized in that, All N electrode segments are of equal length.

6. The electro-optic modulator bandwidth equalization structure according to claim 1, characterized in that, The lengths of the N electrode segments gradually decrease along the signal transmission direction, and the ratio of the decrease in length of each electrode segment is equal to the ratio of the decrease in the active region load ratio within the corresponding modulation segment.

7. The electro-optic modulator bandwidth equalization structure according to claim 1, characterized in that, The electro-optic modulator bandwidth equalization structure also includes a terminal load; When the terminal load is a matched load, the impedance of the terminal load is equal to the characteristic impedance of the traveling wave electrode to achieve optimal signal quality. When the terminal load is a mismatched load, the impedance of the terminal load is lower than the characteristic impedance of the traveling wave electrode, thereby enhancing the high-frequency response of the electro-optic modulator.

8. The electro-optic modulator bandwidth equalization structure according to claim 1, characterized in that, The gradual reduction of the active region load ratio in each modulation section is achieved through distributed bias voltage; wherein, along the signal transmission direction, the reverse bias voltage applied to each modulation section gradually increases, so as to equivalently achieve the sequential reduction of the active region load ratio.

9. The electro-optic modulator bandwidth equalization structure according to claim 1, characterized in that, The active region load is the active region of a silicon photonic modulator or the active region of a III-V compound semiconductor modulator.

10. An electro-optic modulator, characterized in that, The electro-optic modulator includes the electro-optic modulator bandwidth equalization structure as described in any one of claims 1 to 9.

Images