Silicon optical modulator and waveguide core layer thereof

Abstract

This invention discloses a silicon photonic modulator and its waveguide core layer, relating to the field of optical communication technology. The silicon photonic modulator includes a waveguide core layer and semiconductor doped regions disposed therein, characterized in that the semiconductor doped regions are asymmetrically distributed about the geometric center line. Specifically, utilizing the difference in light absorption sensitivity between P-type and N-type doping, the moderately doped P+ regions are made to move more towards the center than the N+ regions, or only the P+ regions are widened towards the center on one side. This structure significantly reduces the microwave loss of the PN junction and significantly improves the photoelectric response bandwidth without significantly increasing optical loss.

Description

Technical Field

[0001] This invention relates to the field of optical communication technology, and in particular to a silicon optical modulator and its waveguide core layer. Background Technology

[0002] Silicon photonic modulators, as core active devices in optical communication and optical interconnect systems, play a crucial role in converting high-speed electrical signals and loading them onto optical carriers, serving as the physical hub for achieving efficient electro-optical conversion. In practical applications, silicon photonic modulators are widely used due to their compatibility with CMOS processes and high integration density. However, with the ever-increasing demands for data transmission rates and bandwidth in modern communications, further improving the modulation efficiency, extinction ratio, and reducing drive power of silicon photonic modulators has become a critical technological challenge that urgently needs to be addressed in this field.

[0003] like Figure 1 As shown, existing silicon photonic modulators generally employ a planar traveling wave electrode structure. From a cross-sectional perspective, the device comprises, from bottom to top, a silicon substrate layer 10, a buried oxide layer 20, and a waveguide core layer 30.

[0004] Specifically, the bottom silicon substrate layer 10 primarily provides robust physical support for the entire chip; immediately above it is the buried oxide layer 20, which utilizes its material properties to vertically confine the light field, preventing optical signals from leaking to the bottom substrate; the waveguide core layer 30 is the core functional area of ​​the entire device, where optical signals are transmitted. This area also contains semiconductor doped regions (such as P-regions and N-regions), where external radio frequency electric fields interact with the optical signals to achieve phase modulation. V1 and the trapezoidal structure connected to the semiconductor doped regions are vias, and V2 is also a via; the rectangular structures M1 and M2 connected to the vias are metal wiring layers.

[0005] like Figure 2 As shown, symmetrical semiconductor doping structures are commonly used in the waveguide core layer design of conventional silicon photonic modulators. Specifically, with the geometric center line of the waveguide as the axis of symmetry, lightly doped P-regions and N-regions are distributed on both sides, followed by moderately doped P+ and N+ regions as transitions, and heavily doped P++ and N++ regions located on the outermost side for connecting electrodes. This symmetrical structure was easy to standardize in early silicon photonic chip designs and could maintain basic photoelectric conversion functionality at a certain modulation rate. However, as optical communication networks evolve towards ultra-high speeds and ultra-large capacities, the system places more stringent requirements on the photoelectric bandwidth of silicon photonic modulators, and the performance bottleneck of existing symmetrical doping structures has gradually become apparent.

[0006] Under the aforementioned symmetrical doping architecture, further improving the bandwidth of silicon photonic modulators faces an irreconcilable physical contradiction. First, during dynamic modulation, when an external driving RF signal is applied to the PN junction, the drastic carrier concentration difference between the heavily doped region (P++ or N++) and the central lightly doped region (P or N) generates a significant potential difference and voltage drop at the semiconductor interface. For example, a nominal 1-volt input driving voltage might lose 0.2 volts at this point. This ineffective voltage dissipation essentially stems from the extremely high intrinsic RF loss within the PN junction.

[0007] To reduce microwave and radio frequency losses and increase bandwidth, an intuitive solution is to widen the P+ and N+ regions on both sides towards the waveguide geometry centerline, replacing some of the original lightly doped regions and thus shortening the transmission dissipation path of the radio frequency signal. However, this traditional symmetrical indentation method towards the center introduces a fatal technical flaw: a sharp increase in optical absorption loss.

[0008] Therefore, existing technologies cannot effectively reduce radio frequency losses without significantly increasing optical absorption losses through simple symmetrical structure adjustments. This has become a key technical challenge restricting the bandwidth of silicon photonic modulators from reaching higher frequency bands. Summary of the Invention

[0009] To address the technical challenge of extending the bandwidth of silicon photonic modulators to higher frequency bands, this invention provides the following technical solution.

[0010] According to a first aspect of the present invention, a waveguide core layer for a silicon photonic modulator is provided, extending along the direction of optical signal transmission, comprising: a cross-section of the waveguide core layer having a geometric center line; a semiconductor doped region disposed within the waveguide core layer and forming a PN junction, the semiconductor doped region comprising at least lightly doped P-regions and lightly doped N-regions distributed on both sides of the geometric center line, and moderately doped P+ regions and moderately doped N+ regions located outside the lightly doped P-regions and lightly doped N-regions respectively; wherein the semiconductor doped region is asymmetrically distributed about the geometric center line: the moderately doped P+ regions widen towards the geometric center line.

[0011] Furthermore, the asymmetric distribution specifically means that both the moderately doped P+ region and the moderately doped N+ region widen towards the geometric center line, and the distance by which the moderately doped P+ region approaches the geometric center line is greater than the distance by which the moderately doped N+ region approaches the geometric center line.

[0012] Furthermore, the asymmetric distribution specifically means that only the moderately doped P+ region widens towards the geometric center line, the boundary line between the lightly doped P region and the moderately doped P+ region moves towards the geometric center line, and the physical width of the lightly doped N region remains unchanged.

[0013] Furthermore, the semiconductor doping region also includes a heavily doped P++ region and a heavily doped N++ region located outside the moderately doped P+ region and the moderately doped N+ region, respectively; the outer boundaries of the heavily doped P++ region and the heavily doped N++ region remain fixed.

[0014] According to a second aspect of the present invention, a silicon optical modulator is provided, comprising the waveguide core layer described in any of the preceding claims.

[0015] Furthermore, it also includes: a traveling wave electrode, located above the waveguide core layer, for receiving externally fed radio frequency signals; the traveling wave electrode adopts a ground-signal-ground coplanar waveguide structure, including a central signal line and ground lines symmetrically distributed on both sides of the central signal line.

[0016] Furthermore, along the direction of radio frequency signal transmission, the traveling wave electrode includes at least a continuous first segment; within the first segment, the center signal line is configured to have a constant initial width.

[0017] Furthermore, along the direction of radio frequency signal transmission, the traveling wave electrode is continuously connected to a second segment after the first segment; in the second segment, the physical width of the center signal line gradually increases along the transmission direction, while the ground line remains straight, so that the local spacing between the center signal line and the ground line is passively compressed and narrowed, thereby maintaining the local electric field strength after the radio frequency signal attenuation.

[0018] Furthermore, the physical width of the center signal line gradually increases linearly along the transmission direction.

[0019] Furthermore, the physical width of the center signal line Coordinates along the transmission direction The change of satisfies the following formula: when hour:

[0020] when hour:

[0021] in, The length of the first segment, The length of the second segment. The initial width used to improve the characteristic impedance of the front end and satisfying , The slope coefficient for the outward widening of the center signal line.

[0022] The present invention has the following technical effects: The core technological advantage of asymmetric doping distribution structures in semiconductors lies in breaking the traditional physical impasse between increasing RF bandwidth and controlling optical absorption loss in silicon photonic modulators. This invention cleverly utilizes the inherent physical difference of the relatively weak optical absorption effect of P-type ions. It widens the moderately doped P+ region significantly more towards the geometric center line than the moderately doped N+ region, or widens only one side of the P+ region. This asymmetric design strategy, under the extremely safe condition of almost no increase in overall optical absorption loss, significantly shortens the physical width of the lightly doped central region, thereby greatly reducing the potential drop and intrinsic RF loss at the PN junction interface. Macroscopically, this directly removes the bottleneck for high-speed signal transmission, resulting in a leapfrog improvement in the device's response bandwidth.

[0023] Due to the significant reduction in microwave loss and cascade resistance of the underlying PN junction, the overall reference impedance of the modulator inevitably decreases. If the RF signal is directly fed in according to conventional design, severe microwave reflection will inevitably occur. This invention can passively increase the local relative spacing by intentionally designing the initial width W1 of the center signal line in the first segment of signal feeding to be smaller than the reference width under standard matching conditions, while keeping the ground lines on both sides fixed. This geometric adjustment effectively improves the inductance per unit length of the front-end traveling wave electrode, precisely raising the originally dropped characteristic impedance and re-anchoring it to the ideal matching state, ensuring efficient, lossless, and reflection-free initial injection of high-frequency microwave signals at the system level.

[0024] The gradually widening central signal line further maximizes the electro-optic modulation efficiency across the entire link. After addressing the issues of front-end impedance matching and overall RF bandwidth, and considering the severe voltage attenuation that inevitably occurs in the latter half of high-speed signal transmission over long distances, this structure passively and smoothly compresses the local spacing between the central signal line and the two flat ground lines by continuously widening the central signal line outwards along the transmission direction. This physical space compression design forcibly amplifies the intensity of the alternating electric field that can be excited locally at the electrode ends where the RF voltage is already very weak. It reactivates the latter half of the waveguide, which had previously become an ineffective dissipation region due to signal attenuation, not only completely extracting residual RF energy but also significantly improving the equivalent modulation efficiency and signal extinction ratio within the same chip size, ultimately achieving comprehensive optimization of the silicon photonic modulator's overall performance. Attached Figure Description

[0025] Figure 1This is a cross-sectional schematic diagram of a silicon photonic modulator in the prior art; Figure 2 This is a schematic diagram of a waveguide core layer in the prior art (only one PN junction is shown). Figure 3 This is a schematic diagram of the waveguide core structure according to an embodiment of the present invention (only one PN junction is shown). Figure 4 This is a performance comparison diagram of the present invention; Figure 5 This is a schematic diagram of the electrode structure according to another embodiment of the present invention (only the electrode structure is shown). Figure reference numerals: 10, silicon substrate layer; 20, buried oxide layer; 30, waveguide core layer; 51, center signal line; 52, first ground line; 53, second ground line; 60, geometric center line of the waveguide. Detailed Implementation

[0026] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0027] like Figure 1 and Figure 2 As shown, in the prior art, a silicon photonic modulator, from a cross-sectional perspective, comprises a silicon substrate layer 10, a buried oxide layer 20, and a waveguide core layer 30, from bottom to top. This invention primarily relates to the waveguide core layer 30 and not to the other layers, therefore they will not be described in detail.

[0028] According to the background technology, the drawback of this structure is that existing technologies cannot effectively reduce radio frequency losses through simple symmetrical structure adjustments without significantly increasing light absorption losses. This has become a key technical challenge restricting the bandwidth of silicon photonic modulators from reaching higher frequency bands.

[0029] The inventors discovered through research that, in conventional solutions, widening both the P+ and N+ regions on both sides towards the waveguide's geometric centerline inevitably leads to absorption of the transmitted optical signal by the doped ions in the waveguide core. Furthermore, there are significant inherent physical differences in the phase changes and absorption degrees caused by P-type and N-type ions. N-type doping is particularly sensitive to optical field absorption; if the N+ region is moved too close to the waveguide's geometric centerline to reduce RF loss, it will immediately cause a severe degradation in the overall modulator's optical loss performance.

[0030] In other words, under the traditional symmetrical design concept, in order to maintain the symmetry of both sides and ensure the optical performance of the N+ region, the P+ region is also restricted to a position far from the geometric center line. This means that in order to accommodate the high absorption characteristics of N-type ions, the design space for further reducing microwave loss through the P+ region was abandoned.

[0031] Based on this, the present invention completely overturns the traditional symmetrical design thinking. The present invention provides the following implementation methods.

[0032] To further improve the electro-optic bandwidth of silicon photonic modulators without significantly increasing optical absorption loss, embodiments of the present invention provide a waveguide core layer design based on an asymmetric doped structure. From the cross-sectional structure of this embodiment, the device also includes, from bottom to top, a silicon substrate layer, a buried oxide layer, a waveguide core layer, an isolation layer, and a top electrode structure.

[0033] The core improvement lies in the doping distribution of the PN junction within the waveguide core layer. In traditional symmetrical designs, the heavily doped regions (P++ and N++), the moderately doped transition regions (P+ and N+), and the lightly doped regions (P and N) are typically mirror-symmetrically distributed along the waveguide's geometric center line 60°. However, in this embodiment, the waveguide core layer employs an asymmetric doping configuration, specifically: The original P+ and N+ regions are widened towards the geometric center line of the waveguide (i.e., the physical width of the P and N regions is relatively shortened), and the P+ region moves towards the center to a much greater extent than the N+ region moves towards the center. Or: such as Figure 3 As shown (the dashed box shows a conventional symmetrical structure), the original P+ region is widened towards the geometric center line 60 of the waveguide (i.e., the central depletion layer of the PN junction). (The boundary between P and P+ is moved towards the geometric center line 60 of the waveguide, which means the physical width of the P region is relatively shortened.)

[0034] The physical mechanism of this asymmetric doping design stems from the inherent differences in conductivity and light absorption between P-type and N-type ions in semiconductors. When the modulator is operating, if an external driving voltage is applied directly between heavily doped and lightly doped regions with vastly different concentrations, a significant potential difference and voltage drop will occur (e.g., a 1-volt input voltage may result in a 0.2-volt loss at the interface). The existence of the P+ and N+ transition regions is precisely to smooth the concentration gradient and reduce the voltage drop. By further widening the P+ and N+ regions towards the waveguide center, they can effectively replace part of the original lightly doped region.

[0035] This physical deformation significantly reduces microwave loss (or radio frequency loss) of alternating drive signals in the PN junction region from the perspective of high-frequency electromagnetics, and simultaneously reduces the equivalent capacitance of the relevant region. The dual reduction in microwave loss and parasitic capacitance is key to overcoming the bottleneck of high-speed signals and improving the overall bandwidth of the modulator.

[0036] Widening the doped region towards the center inevitably introduces side effects, namely, the dopant ions generate additional absorption loss for the optical signal transmitted in the waveguide. Since N-type doping is extremely sensitive to light absorption, if the N+ region is too close to the geometric center line of the waveguide, the modulator's optical loss performance will deteriorate drastically. Therefore, the N+ region can only be extended inwards as much as possible within a safe distance that does not significantly increase light absorption. In contrast, P-type doping has a relatively weaker light absorption effect, which gives the P+ region greater design redundancy. It can be moved closer to the geometric center line than the N+ region without causing a significant increase in optical loss. Based on this asymmetry in material properties, this embodiment specifically employs an asymmetric doping pattern in the design of the photomask, maximizing the benefits of reducing RF loss while minimizing the impact on light absorption.

[0037] like Figure 4 As shown, taking specific test data comparison as an example, this asymmetric doping structure exhibits extremely excellent technical performance in reducing microwave loss. Combined with the microwave loss comparison graph before and after improvement provided in the embodiment, it can be seen that the horizontal axis of the graph represents the operating frequency of the RF signal, and the vertical axis represents the microwave loss magnitude (unit: dB / mm). Throughout the entire high-frequency operating range, after adopting the asymmetric inward widening design of the P+ and N+ regions of this embodiment (i.e., the "after improvement" curve in the figure), its microwave loss is significantly lower than that of the conventional design (i.e., the "before improvement" curve in the figure), and this advantage of reduced microwave loss becomes increasingly apparent as the operating frequency continues to increase.

[0038] This significant reduction in high-frequency microwave loss directly eliminates the physical bottleneck restricting high-speed signal transmission, thus successfully translating into a substantial improvement in the overall photoelectric response bandwidth of the device. According to actual test calculations, in the conventional design before the improvement, the -3dB photoelectric bandwidth of this silicon photonic modulator was approximately 41GHz; while with the reduction in microwave loss… Figure 4 The significant improvement shown indicates that the modulator's -3dB optoelectronic bandwidth was successfully extended to 52GHz. This means that the bandwidth performance was improved by approximately 25% (an increase of over 10GHz) with almost no change to the overall physical dimensions of the device. More importantly, due to strict control of the N+ region boundary and full utilization of the low absorption characteristics of the P+ region, the cost of achieving this significant reduction in microwave loss and bandwidth gain was minimal. The additional optical absorption loss of the modulator as a whole was perfectly controlled within 0.5dB, successfully achieving the optimal physical balance between reducing microwave loss and controlling optical loss.

[0039] The shortening of the distance between the internal doped regions inevitably leads to a decrease in the overall reference impedance of the modulator, posing a severe impedance mismatch challenge for the non-reflective feeding of the front-end RF signal.

[0040] To offset the side effects of the aforementioned impedance reduction and to simultaneously address the issue of low modulation efficiency in the latter half of the RF signal due to voltage attenuation, in another embodiment, improvements are made to the GSG planar traveling wave electrode above the waveguide core layer.

[0041] like Figure 5 As shown, the radio frequency signal and the optical signal are transmitted in the same direction. In the axial direction, the traveling wave electrode is clearly divided into the first half. and the second half Two functional areas. In the first half of the interval, the first ground line 52 and the second ground line 53 on both sides remain parallel, and the center signal line 51 has a constant initial width. Crucially, this initial width is designed to compensate for the low impedance effect caused by the asymmetric doping in the underlying layer. Intentionally designed to be significantly smaller than the reference width of conventional standard electrodes (Right now , (This is the reference width in existing technology). Based on the impedance distribution characteristics of microwave transmission lines, with the physical positions of the ground wires on both sides fixed, the width of the center signal line... The reduction of the local spacing is equivalent to increasing the relative proportion of the local spacing, which can effectively improve the inductance per unit length and reduce the capacitance, ensuring the efficient and reflection-free injection of the initial high-frequency microwave signal.

[0042] When the radio frequency signal is transmitted smoothly into In the latter half of the interval, the voltage amplitude will inevitably decrease significantly due to various physical dissipations along the path. At this time, the ground lines on both sides still extend straight without deflection, but the physical width of the center signal line... It begins by gradually widening outwards along the transmission direction. This gradual widening follows a piecewise linear function: ,in This is the slope coefficient for the center signal line to widen outwards. As the center signal line... As the segment widens, the local distance between it and the ground wires on both sides is passively and gradually compressed. This continuous compression of physical space forcibly amplifies the local electric field intensity at the electrode ends, where the radio frequency voltage is already very weak. This fully converts the previously unusable residual radio frequency energy into effective electro-optic phase shift.

[0043] In summary, this embodiment endows the device with extremely high response bandwidth through bottom-layer asymmetric doping, while cleverly utilizing the geometric deformation of the top-layer center signal line—"first narrowing the matching, then widening and extracting"—to correct impedance mismatch at the front end and significantly increase the equivalent modulation efficiency at the end, achieving a comprehensive leap in the overall performance of the silicon photonic modulator.

[0044] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A waveguide core layer for use in a silicon optical modulator, extending along the direction of optical signal transmission, wherein the cross-section of the waveguide core layer has a geometric centerline, characterized in that, include: A semiconductor doped region is disposed within the waveguide core layer and forms a PN junction. The semiconductor doped region includes at least a lightly doped P region and a lightly doped N region distributed on both sides of the geometric center line, and a moderately doped P+ region and a moderately doped N+ region located outside the lightly doped P region and the lightly doped N region, respectively. The semiconductor doped regions are asymmetrically distributed about the geometric center line: the moderately doped P+ regions widen towards the geometric center line; specifically, the asymmetrical distribution means that both the moderately doped P+ regions and the moderately doped N+ regions widen towards the geometric center line, and the distance by which the moderately doped P+ regions approach the geometric center line is greater than the distance by which the moderately doped N+ regions approach the geometric center line.

2. The waveguide core layer according to claim 1, characterized in that, The semiconductor doping region also includes a heavily doped P++ region and a heavily doped N++ region located outside the moderately doped P+ region and the moderately doped N+ region, respectively; the outer boundaries of the heavily doped P++ region and the heavily doped N++ region remain fixed.

3. A silicon photonic modulator, characterized in that, Includes the waveguide core layer as described in any one of claims 1 to 2.

4. The silicon photonic modulator according to claim 3, characterized in that, Also includes: A traveling wave electrode, located above the waveguide core layer, is used to receive externally fed radio frequency signals. The traveling wave electrode adopts a ground-signal-ground coplanar waveguide structure, including a central signal line and ground lines symmetrically distributed on both sides of the central signal line.

5. The silicon photonic modulator according to claim 4, characterized in that, Along the direction of radio frequency signal transmission, the traveling wave electrode includes at least a continuous first segment; within the first segment, the center signal line is configured to have a constant initial width.

6. The silicon photonic modulator according to claim 5, characterized in that, Along the direction of radio frequency signal transmission, the traveling wave electrode is continuously connected to a second segment after the first segment; in the second segment, the physical width of the center signal line gradually increases along the transmission direction, while the ground line remains straight, so that the local spacing between the center signal line and the ground line is passively compressed and narrowed, thereby maintaining the local electric field strength after the radio frequency signal attenuation.

7. The silicon photonic modulator according to claim 6, characterized in that, The physical width of the center signal line gradually increases linearly along the transmission direction.

8. The silicon photonic modulator according to claim 7, characterized in that, The physical width of the center signal line Coordinates along the transmission direction The change of satisfies the following formula: when hour: when hour: in, The length of the first segment, The length of the second segment. The initial width used to improve the characteristic impedance of the front end and satisfying , The slope coefficient for the outward widening of the center signal line.

Images