CN114325931A - Manufacturing method of silicon optical device, silicon optical device and photonic integrated circuit - Google Patents

Manufacturing method of silicon optical device, silicon optical device and photonic integrated circuit Download PDF

Info

Publication number
CN114325931A
CN114325931A CN202111590901.7A CN202111590901A CN114325931A CN 114325931 A CN114325931 A CN 114325931A CN 202111590901 A CN202111590901 A CN 202111590901A CN 114325931 A CN114325931 A CN 114325931A
Authority
CN
China
Prior art keywords
optical device
silicon optical
dimensional code
input
rectangular module
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202111590901.7A
Other languages
Chinese (zh)
Other versions
CN114325931B (en
Inventor
付红岩
毛思梅
程理荣
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen International Graduate School of Tsinghua University
Original Assignee
Shenzhen International Graduate School of Tsinghua University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenzhen International Graduate School of Tsinghua University filed Critical Shenzhen International Graduate School of Tsinghua University
Priority to CN202111590901.7A priority Critical patent/CN114325931B/en
Publication of CN114325931A publication Critical patent/CN114325931A/en
Application granted granted Critical
Publication of CN114325931B publication Critical patent/CN114325931B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02PCLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P70/00Climate change mitigation technologies in the production process for final industrial or consumer products
    • Y02P70/50Manufacturing or production processes characterised by the final manufactured product

Landscapes

  • Optical Integrated Circuits (AREA)

Abstract

The application discloses silicon optical device's manufacturing approach, silicon optical device and photon integrated circuit, silicon optical device include two input waveguides, two output waveguides and the rectangle module that comprises a plurality of pixel block, and input waveguide sets up in the both sides of rectangle module with output waveguide relatively, and the rectangle module is used for handling the signal of being input by input waveguide, and wherein, silicon optical device's manufacturing approach includes: acquiring a target transmission matrix; inputting the target transmission matrix into a structure generation model for prediction to obtain a two-dimensional code structure; and etching the rectangular module according to the two-dimensional code structure. According to the embodiment of the application, the two-dimension code structure corresponding to the target transmission matrix is obtained through the structure generation model, and the rectangular module is etched according to the two-dimension code structure, so that the silicon optical device with a specific function is obtained, meanwhile, the time and resources spent by the analog device are reduced, and the design and manufacturing period of the silicon optical device can be effectively shortened.

Description

Manufacturing method of silicon optical device, silicon optical device and photonic integrated circuit
Technical Field
The present disclosure relates to the field of optical devices, and particularly to a method for manufacturing a silicon optical device, and a photonic integrated circuit.
Background
With the progress of science and technology, the processing demand of big data is rapidly increased year by year, and the problems of large energy consumption and low data transmission efficiency of an integrated circuit can be solved by the concept of optical charge of a photonic integrated circuit, so that the photonic integrated circuit is widely applied to the field of optical communication.
However, in the related art, the photonic integrated circuit has a complex design, and the device structures with different functions have large differences, so that a large amount of time and resources are required to simulate to set a device with a specific function.
Disclosure of Invention
The present application is directed to solving at least one of the problems in the prior art. Therefore, the application provides a manufacturing method of a silicon optical device, which can improve the processing speed of the device and save the processing time.
According to the manufacturing method of the silicon optical device in the embodiment of the first aspect of the present application, the silicon optical device includes two input waveguides, two output waveguides and a rectangular module composed of a plurality of pixel blocks, the input waveguides and the output waveguides are oppositely arranged on two sides of the rectangular module, and the rectangular module is used for processing signals input by the input waveguides;
the manufacturing method of the silicon optical device comprises the following steps:
acquiring a target transmission matrix;
inputting the target transmission matrix into a structure generation model for prediction to obtain a two-dimensional code structure;
and etching the rectangular module according to the two-dimension code structure.
According to the manufacturing method of the silicon optical device, the following beneficial effects are at least achieved: the two-dimensional code structure corresponding to the target transmission matrix is obtained through the structure generation model, and the rectangular module is etched according to the two-dimensional code structure, so that the silicon optical device with a specific function is obtained, meanwhile, the time and resources spent by the analog device are reduced, and the design and manufacturing period of the silicon optical device can be effectively shortened.
According to some embodiments of the present application, before inputting the target transmission matrix into a structure generation model for prediction to obtain a two-dimensional code structure, the method further includes generating the structure generation model, specifically including: predicting the predicted two-dimensional code structure through a pre-training model to obtain a predicted transmission matrix; obtaining a loss function according to a target transmission matrix and the prediction transmission matrix; calculating the loss function to obtain a loss value; and training the pre-training model according to the loss value to obtain the structure generation model.
According to some embodiments of the present application, the training the pre-trained model according to the loss value to obtain a structure generation model includes: taking the loss value as a reverse propagation quantity, and adjusting model parameters of the pre-training model to train the pre-training model to obtain the structure generation model; wherein the model parameters include weight values.
According to some embodiments of the present application, the etching the rectangular module according to the two-dimensional code structure includes: acquiring the two-dimension code structure; and etching the pixel blocks corresponding to the rectangular modules according to the two-dimensional code structure.
According to some embodiments of the present application, the etching the pixel block corresponding to the rectangular module according to the two-dimensional code structure includes: if the code value of the two-dimensional code structure is 1, etching the corresponding pixel block to form a groove, and filling the groove; and if the code value of the two-dimensional code structure is 0, the corresponding pixel block is not etched.
A silicon optical device according to an embodiment of the second aspect of the present application includes: a rectangular module comprising a plurality of pixel blocks; the rectangular module is used for processing an input signal, and a groove is formed in the center of the pixel block, corresponding to the two-dimensional code structure, of the rectangular module; two input waveguides, connected to one side of the rectangular module, for receiving the input signal; and the output waveguides are connected with one side of the rectangular module, which is far away from the input waveguides, and are used for outputting the input signals processed by the rectangular module.
According to the silicon optical device of the embodiment of the application, the following beneficial effects are at least achieved: by arranging the rectangular module, the input signal can be processed, so that a target signal is obtained, the transmission performance of the silicon optical device is improved, the size of the silicon optical device is reduced, and the silicon optical device is convenient to apply practically.
According to some embodiments of the present application, the recess is cylindrical, and the recess is filled with a silicon dioxide material therein.
According to some embodiments of the application, the grooves have a diameter of 90 nm.
According to some embodiments of the present application, the rectangular module is 3 μm long and 3 μm wide, and the rectangular module is composed of 25 identical pixel blocks.
A photonic integrated circuit according to an embodiment of the second aspect of the present application comprises a silicon optical device according to an embodiment of the second aspect of the present application described above.
The photonic integrated circuit according to the embodiment of the application has at least the following beneficial effects: through setting up above-mentioned silicon optical device, can improve signal transmission's speed, reduce the consumption of the energy, can realize different functions according to the demand simultaneously, improve photonic integrated circuit's flexibility.
Additional aspects and advantages of the present application will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the present application.
Drawings
The present application is further described with reference to the following figures and examples, in which:
FIG. 1 is a schematic flow chart illustrating a method for fabricating a silicon optical device according to an embodiment of the present disclosure;
FIG. 2 is a schematic structural diagram of a silicon optical device according to an embodiment of the present application;
3 a-3 d are schematic structural diagrams of a silicon optical device according to an embodiment of the present application;
FIG. 4 is a schematic flow chart illustrating a method for fabricating a silicon optical device according to an embodiment of the present application;
FIG. 5 is a schematic diagram of the detailed process of step S300 in FIG. 1;
FIG. 6 is a flowchart illustrating the detailed process of step S320 in FIG. 5;
fig. 7 is a schematic application diagram of a silicon optical device according to an embodiment of the present application.
Reference numerals:
an input waveguide 100, a first input waveguide 110, a second input waveguide 120, an output waveguide 200, a first output waveguide 210, a second output waveguide 220, a rectangular module 300, a block of pixels 310, a groove 311.
Detailed Description
Reference will now be made in detail to embodiments of the present application, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the drawings are exemplary only for the purpose of explaining the present application and are not to be construed as limiting the present application.
In the description of the present application, it is to be understood that the positional descriptions, such as the directions of up, down, front, rear, left, right, etc., referred to herein are based on the directions or positional relationships shown in the drawings, and are only for convenience of description and simplification of description, and do not indicate or imply that the referred device or element must have a specific direction, be constructed and operated in a specific direction, and thus, should not be construed as limiting the present application.
In the description of the present application, the meaning of a plurality is one or more, the meaning of a plurality is two or more, and the above, below, exceeding, etc. are understood as excluding the present number, and the above, below, within, etc. are understood as including the present number. If the first and second are described for the purpose of distinguishing technical features, they are not to be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated or implicitly indicating the precedence of the technical features indicated.
In the description of the present application, unless otherwise expressly limited, terms such as set, mounted, connected and the like should be construed broadly, and those skilled in the art can reasonably determine the specific meaning of the terms in the present application by combining the detailed contents of the technical solutions.
In the description of the present application, reference to the description of the terms "one embodiment," "some embodiments," "an illustrative embodiment," "an example," "a specific example," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present application. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
A method of manufacturing a silicon optical device according to an embodiment of the present application is described below with reference to fig. 1 to 3 d.
As shown in fig. 1 and 2, the silicon optical device includes two input waveguides 100, two output waveguides 200, and a rectangular module 300 composed of a plurality of pixel blocks 310, the input waveguides 100 are disposed on both sides of the rectangular module 300 opposite to the output waveguides 200, and the rectangular module 300 is used for processing signals input by the input waveguides 100.
The manufacturing method of the silicon optical device comprises the following steps of:
step S100: acquiring a target transmission matrix;
step S200: inputting the target transmission matrix into a structure generation model for prediction to obtain a two-dimensional code structure;
step S300: and etching the rectangular module according to the two-dimensional code structure.
Specifically, as shown in fig. 2, the silicon optical device includes an input waveguide 100, an output waveguide 200, and a rectangular module 300. The rectangular module 300 has a substantially plate-like structure, and is composed of a plurality of pixel blocks 310 with the same size and shape, and when a signal passes through the rectangular module 300, linear calculation is performed, so that a signal required by a user is output. The two input waveguides 100 are defined as a first input waveguide 110 and a second input waveguide 120, respectively, and the two output waveguides 200 are defined as a first output waveguide 210 and a second output waveguide 220, respectively. Optical signals are input into the silicon optical device from the first input waveguide 110 and the second input waveguide 120, and after interaction with the rectangular module 300, modulated optical signals are output from the first output waveguide 210 and the second output waveguide 220.
By changing the rectangular module 300, the output signal can be modulated. The regulation and control of the matrix module on the phase and the copy of the input signal are defined as a transmission matrix T:
Figure BDA0003429060830000041
wherein, amnIs the amplitude variation factor of the input signal from the input waveguide 100 to the output waveguide 200,
Figure BDA0003429060830000042
Is the phase shift value of the input signal from the input waveguide 100 to the output waveguide 200.
Define the input signal of the first input waveguide 110 as EI1The input signal of the second input waveguide 120 is defined as EI2The output signal of the first output waveguide 210 is defined as EO1The output signal of the second output waveguide 220 is defined as EO2Then input signal EI1And an output signal EI2The relationship of (1) is:
Figure BDA0003429060830000043
the user can change the transmission matrix T of the transmission module by changing the etching condition of the pixel block 310 in the rectangular module 300, thereby realizing the transmission and operation of the input signal and obtaining the required signal.
As shown in fig. 1, when a silicon optical device is manufactured, a target transmission matrix is obtained according to a user signal processing requirement, and the target transmission matrix is input into a structure generation model, where a corresponding relationship between the transmission matrix and a two-dimensional code structure is stored in the structure generation model. And predicting the structure generation model according to the target transmission matrix to obtain a corresponding two-dimensional code structure, and etching the rectangular module 300 according to the two-dimensional code structure to obtain the target silicon optical device.
For example, when the silicon optical device is applied as a crossed waveguide, the target transmission matrix
Figure BDA0003429060830000051
Inputting the target transmission matrix into the structure generation model to obtain a corresponding two-dimensional code structure, and etching the region corresponding to the rectangular module 300 according to the two-dimensional code structure to obtain a two-dimensional code pattern as shown in fig. 3 a. The silicon optical device realizes signal exchange after signals input from the input waveguide pass through the rectangular module 300, that is, signals input from the first input waveguide 110 are output by the second output waveguide 220, and signals input from the second input waveguide 120 are output by the first output waveguide 210.
When the silicon optical device is used as a phase shifter, the target transmission matrix
Figure BDA0003429060830000052
Inputting the target transmission matrix into the structure generation model to obtain a corresponding two-dimensional code structure, and etching the region corresponding to the rectangular module 300 according to the two-dimensional code structure to obtain a two-dimensional code pattern as shown in fig. 3 b. In the silicon optical device, after a signal input from the input waveguide 100 passes through the rectangular module 300, the amplitude and the transmission path of the signal are kept unchanged, but the phase of the signal output from the output waveguide 200 is changed.
When the silicon optical device is used as an exclusive-OR gate, a target transmission matrix
Figure BDA0003429060830000053
Inputting the target transmission matrix into the structure generation model to obtain a corresponding two-dimensional code structure, and etching the region corresponding to the rectangular module 300 according to the two-dimensional code structure to obtain a two-dimensional code pattern as shown in fig. 3 c. When the input signal of the first input waveguide 110 is 1 and the input signal of the second input waveguide 120 is 0, or the input signal of the first input waveguide 110 is 0 and the input signal of the second input waveguide 120 is 1, the signal output by the first output waveguide 210 is 1; when the input signals of the first input waveguide 110 and the second input waveguide 120 are both 0 or both 1, the signal output by the first output waveguide 210 is 0.
Target transmission when applying silicon optical device as random unitary matrix coreMatrix array
Figure BDA0003429060830000054
Inputting the target transmission matrix into the structure generation model to obtain a corresponding two-dimensional code structure, and etching the region corresponding to the rectangular module 300 according to the two-dimensional code structure to obtain a two-dimensional code pattern as shown in fig. 3 d. When the input signal passes through the silicon optical matrix, interference occurs, and the amplitude and the phase of the output light are modulated.
According to the manufacturing method of the silicon optical device, the two-dimensional code structure corresponding to the target transmission matrix is obtained through the structure generation model, and the rectangular module 300 is etched according to the two-dimensional code structure, so that the silicon optical device with the specific function is obtained, meanwhile, time and resources spent by a simulation device are reduced, and the design and manufacturing period of the silicon optical device can be effectively shortened.
In some embodiments of the present application, as shown in fig. 4, the method of fabricating a silicon optical device further comprises generating a structure generation model, including but not limited to the steps of:
step S400: predicting the predicted two-dimensional code structure through a pre-training model to obtain a predicted transmission matrix;
step S500: obtaining a loss function according to the target transmission matrix and the prediction transmission matrix;
step S600: a loss function is calculated to obtain a loss value.
Step S700: and training the pre-training model according to the loss value to obtain a structure generation model.
Specifically, the pre-training model stores some corresponding relations between the transmission matrix and the two-dimensional code structure, and the pre-training model obtains a predicted two-dimensional code structure according to the target transmission matrix and performs prediction processing on the predicted two-dimensional code structure, so as to obtain a predicted transmission matrix. And obtaining the average absolute error value of the target transmission matrix and the prediction transmission matrix so as to obtain a loss function. And correspondingly calculating the loss function to obtain a loss value. And finally, training the pre-training model according to the loss value to enable the pre-training model to be optimized towards a new target, so that the pre-training model is trained by adjusting the parameters of the pre-training model to obtain a trained structure generation model, wherein the structure generation model is used for generating a corresponding two-dimensional code structure according to the target transmission matrix.
Through the trained structure generation model, a user can provide a corresponding two-dimensional code structure according to any specified target transmission matrix without other additional numerical values, and the design efficiency of the silicon optical device can be effectively improved; and the structure generation model has stronger generalization capability and does not need to prepare huge data samples in advance.
In some embodiments of the present application, step S700: training the pre-training model according to the loss value to obtain a structure generation model, comprising: adjusting model parameters of the pre-training model by taking the loss value as a reverse propagation quantity so as to train the pre-training model to obtain a structure generation model; wherein the model parameters include weight values.
Specifically, after a loss function is obtained according to the target transmission matrix and the predicted transmission matrix, derivative calculation is performed on the loss function, so that a loss value is obtained.
Figure BDA0003429060830000061
Wherein,
Figure BDA0003429060830000062
and the value is a loss value, L is a loss function, T' is a prediction transmission matrix, Q is a prediction two-dimensional code structure, and w is a weight value.
Since the loss function L ═ f3(T, T'), wherein T is a target transmission matrix; predicted transmission matrix T ═ f2(Q), predicting two-dimensional code structure Q ═ f1(T) and, therefore,
Figure BDA0003429060830000063
for the explicit function, it can be obtained by the way of partial derivation,
Figure BDA0003429060830000064
which is an implicit function, can be obtained by means of inverse simulation,
Figure BDA0003429060830000065
the explicit function can be obtained by derivation.
Will lose value
Figure BDA0003429060830000071
Adjusting model parameters as a back propagation quantity, wherein the model parameters include a weight value w,
Figure BDA0003429060830000072
where α is a learning rate, and the learning rate α can be set as required.
And adjusting the weight value through the loss value, so that the pre-training model is updated and iterated according to different target transmission matrixes, and a structure generation model is obtained. The training of the structure generation model does not need to prepare a large number of training data samples in advance, and the corresponding two-dimensional code structure can be quickly predicted according to the target transmission matrix after a certain number of iterations, so that the design time of the silicon optical device is shortened.
In some embodiments of the present application, as shown in fig. 5, step S300: the rectangular module 300 is etched according to the two-dimensional code structure, which includes but is not limited to the following steps:
step S310: acquiring a two-dimensional code structure;
step S320: and etching the corresponding pixel block in the rectangular module according to the two-dimensional code structure.
Specifically, after the target transmission matrix is input into the structure generation model, the structure generation model outputs a corresponding two-dimensional code structure. And etching the pixel blocks 310 in the rectangular module 300 according to the two-dimensional code structure, thereby obtaining the rectangular module 300 with the two-dimensional code pattern.
In some embodiments of the present application, as shown in fig. 6, step S320: the etching process is performed on the pixel block 310 corresponding to the rectangular module 300 according to the two-dimensional code structure, and includes, but is not limited to, the following steps:
step S321: if the code value of the two-dimensional code structure is 1, etching the corresponding pixel block to form a groove, and filling the groove;
step S322: and if the code value of the two-dimensional code structure is 0, the corresponding pixel block is not etched.
Specifically, a two-dimensional code structure corresponding to the target transmission matrix is obtained through a structure generation model, wherein the two-dimensional code structure comprises a plurality of characters with code values of 0 or 1. When the code value of the two-dimensional code structure is 1, the pixel block 310 corresponding to the code value in the rectangular module 300 is etched, so that a groove 311 is formed, and the groove 311 is filled, wherein the groove 311 may be a cylindrical groove, and a filling material in the groove 311 may be silicon dioxide. When the code value of the two-dimensional code structure is 0, the pixel block 310 corresponding to the code value in the rectangular module 300 is not etched. Through the above operation, the same pattern as the two-dimensional code structure can be formed on the rectangular module 300. Whether the pixel blocks 310 in the rectangular module 300 are etched or not affects the transmission matrix, so that each two-dimensional code structure corresponds to one transmission matrix, and a corresponding transmission function can be realized by selecting a proper transmission matrix.
In some embodiments, the present application also proposes a silicon optical device, as shown in fig. 2, comprising a rectangular module 300, two input waveguides 100 and two output waveguides 200. The rectangle module 300 includes a plurality of pixel blocks 310; the rectangular module 300 is used for processing an input signal, and a groove 311 is formed in the center of the rectangular module 300 and a pixel block 310 corresponding to the two-dimensional code structure; the input waveguide 100 is connected to one side of the rectangular module 300 for receiving an input signal; the output waveguide 200 is connected to a side of the rectangular module 300 away from the input waveguide 100, and is used for outputting the input signal processed by the rectangular module 300.
Specifically, the rectangular module 300 is substantially a plate-shaped structure and is composed of a plurality of pixel blocks 310 with the same size and shape, and the rectangular module 300 and the pixel block 310 corresponding to the two-dimensional code structure with the code value of 1 are etched, so that a groove 311 is formed in the center of the pixel block 310; the pixel block 310 corresponding to the code value of 0 in the two-dimensional code structure is not etched, so that a pattern corresponding to the two-dimensional code structure is formed on the rectangular module 300. When the signal passes through the rectangular module 300, a corresponding linear calculation is performed, thereby outputting a signal required by a user. The two input waveguides 100 are defined as a first input waveguide 110 and a second input waveguide 120, respectively, and the two output waveguides 200 are defined as a first output waveguide 210 and a second output waveguide 220, respectively. The first input waveguide 110 and the second input waveguide 120 are connected to one side of the rectangular module 300, and the first output waveguide 210 and the second output waveguide 220 are connected to the other side of the rectangular module 300, wherein the first input waveguide 110 and the second input waveguide 120 are disposed opposite to the first output waveguide 210 and the second output waveguide 220. The first input waveguide 110 and the second input waveguide 120 input optical signals into the silicon optical device, and after the optical signals interact with the rectangular module 300, the modulated optical signals are output by the first output waveguide 210 and the second output waveguide 220.
According to the silicon optical device provided by the embodiment of the application, the rectangular module 300 is arranged, so that an input signal can be processed, a target signal is obtained, the transmission performance of the silicon optical device is improved, the size of the silicon optical device is reduced, and the silicon optical device is convenient to use in practice.
In some embodiments of the present application, the recess 311 is cylindrical, and the recess 311 is filled with a silicon dioxide material.
In some embodiments of the present application, the size of the recess 311 is smaller than the size of the pixel block 310. The size of the groove 311 is smaller than the size of a single pixel block 310 in the transmission module, and generally, the difference between the sizes of the groove 311 and the pixel block 310 is greater than or equal to 30nm, so that the processing of the groove 311 can be facilitated, and the processing rate of the silicon optical device can be improved.
In some embodiments of the present application, the pixel block 310 is less than one third of the input wavelength in size. Specifically, the size of the pixel blocks 310 in the transmission module may be set according to the input wavelength of the silicon optical device, the size of the pixel blocks 310 is usually smaller than one third of the input wavelength, and the size of the rectangular module 300 is set according to the number and arrangement of the pixel blocks 310. For example, 25 square pixel blocks 310 each having a side of 120nm are provided, and the pixel blocks 310 are arranged in a 5 × 5 arrangement, thereby obtaining a rectangular module 300 having a length and a width of 3 μm. For convenience of processing, the diameter of the groove 311 formed in the center of the pixel block 310 is 90 nm.
In other embodiments, the 2 × 2-port silicon optical device proposed in the present application may further implement any high-order mxn transmission matrix kernel through concatenation, which can be used for constructing a weight matrix of an optical neural network. For example, a 3 × 3 transmission matrix core is implemented by cascading 32 × 2 ports of silicon optical devices, as shown in fig. 7, the transmission matrix core is:
Figure BDA0003429060830000091
each transmission coefficient in the transmission matrix core can be obtained from target transmission matrixes T (1), T (2) and T (3) of the silicon optical device with three 2 multiplied by 2 ports and a waveguide phase shift value phi corresponding to the device length L. By means of path tracing, any one output port I can be calculatedmTo the output port OnThe transmission coefficient of (1). E.g. by a first input port I1To the first output port O1Has a transmission coefficient of
Figure BDA0003429060830000092
From a first input port I1To the second output port O2Has a transmission coefficient of
Figure BDA0003429060830000093
The rest of the cases are not described herein. Any high-order matrix core can be obtained by cascading a plurality of 2 x 2 port silicon optical devices, the high-order matrix core can be used for replacing a weight matrix of a deep neural network to construct an optical network, and the calculation and operation speed of the optical network can be increased.
In some embodiments, the present application also provides a photonic integrated circuit including the silicon optical device in the above embodiments of the present application.
According to the photonic integrated circuit provided by the embodiment of the application, by arranging the silicon optical device, the signal transmission rate can be improved, the energy consumption can be reduced, different functions can be realized according to requirements, and the flexibility of the photonic integrated circuit is improved.
The embodiments of the present application have been described in detail with reference to the drawings, but the present application is not limited to the embodiments, and various changes can be made within the knowledge of those skilled in the art without departing from the gist of the present application. Furthermore, the embodiments and features of the embodiments of the present application may be combined with each other without conflict.

Claims (10)

1. The manufacturing method of the silicon optical device is characterized in that the silicon optical device comprises two input waveguides, two output waveguides and a rectangular module consisting of a plurality of pixel blocks, wherein the input waveguides and the output waveguides are oppositely arranged on two sides of the rectangular module, and the rectangular module is used for processing signals input by the input waveguides;
the manufacturing method of the silicon optical device comprises the following steps:
acquiring a target transmission matrix;
inputting the target transmission matrix into a structure generation model for prediction to obtain a two-dimensional code structure;
and etching the rectangular module according to the two-dimension code structure.
2. The method according to claim 1, wherein before the target transmission matrix is input into a structure generation model for prediction to obtain the two-dimensional code structure, the method further comprises generating the structure generation model, specifically comprising:
predicting the predicted two-dimensional code structure through a pre-training model to obtain a predicted transmission matrix;
obtaining a loss function according to a target transmission matrix and the prediction transmission matrix;
calculating the loss function to obtain a loss value;
and training the pre-training model according to the loss value to obtain the structure generation model.
3. The method according to claim 2, wherein the training the pre-trained model according to the loss value to obtain a structure generation model comprises:
taking the loss value as a reverse propagation quantity, and adjusting model parameters of the pre-training model to train the pre-training model to obtain the structure generation model; wherein the model parameters include weight values.
4. The method according to claim 1, wherein the etching the rectangular module according to the two-dimensional code structure includes:
acquiring the two-dimension code structure;
and etching the pixel blocks corresponding to the rectangular modules according to the two-dimensional code structure.
5. The method according to claim 4, wherein the etching the pixel block corresponding to the rectangular module according to the two-dimensional code structure includes:
if the code value of the two-dimensional code structure is 1, etching the corresponding pixel block to form a groove, and filling the groove;
and if the code value of the two-dimensional code structure is 0, the corresponding pixel block is not etched.
6. A silicon optical device, comprising:
a rectangular module comprising a plurality of pixel blocks; the rectangular module is used for processing an input signal, and a groove is formed in the center of the pixel block, corresponding to the two-dimensional code structure, of the rectangular module;
two input waveguides, connected to one side of the rectangular module, for receiving the input signal;
and the output waveguides are connected with one side of the rectangular module, which is far away from the input waveguides, and are used for outputting the input signals processed by the rectangular module.
7. The silicon optical device as claimed in claim 6, wherein the recess is cylindrical and filled with a silicon dioxide material.
8. The silicon light device of claim 7, wherein the size of the groove is smaller than the size of the block of pixels.
9. The method of claim 6, wherein the pixel block has a size less than one third of an input wavelength.
10. Photonic integrated circuit, characterized in that it comprises a silicon optical device according to any of claims 6 to 9.
CN202111590901.7A 2021-12-23 2021-12-23 Method for manufacturing silicon optical device, silicon optical device and photonic integrated circuit Active CN114325931B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111590901.7A CN114325931B (en) 2021-12-23 2021-12-23 Method for manufacturing silicon optical device, silicon optical device and photonic integrated circuit

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111590901.7A CN114325931B (en) 2021-12-23 2021-12-23 Method for manufacturing silicon optical device, silicon optical device and photonic integrated circuit

Publications (2)

Publication Number Publication Date
CN114325931A true CN114325931A (en) 2022-04-12
CN114325931B CN114325931B (en) 2023-11-17

Family

ID=81054862

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111590901.7A Active CN114325931B (en) 2021-12-23 2021-12-23 Method for manufacturing silicon optical device, silicon optical device and photonic integrated circuit

Country Status (1)

Country Link
CN (1) CN114325931B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116112088A (en) * 2023-04-13 2023-05-12 中国科学院长春光学精密机械与物理研究所 Adaptive modulation method for multi-input multi-output type photon device

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105334575A (en) * 2015-12-14 2016-02-17 华中科技大学 Silicon-based optical beam splitter and manufacturing method thereof
CN106405730A (en) * 2016-11-28 2017-02-15 华中科技大学 Silicon-based coarse wavelength division device
CN110221384A (en) * 2019-06-17 2019-09-10 华中科技大学 A kind of silicon substrate Meta Materials multimode curved waveguide and preparation method thereof
CN110646883A (en) * 2019-09-30 2020-01-03 华东师范大学重庆研究院 Three-way silicon-based beam splitter chip and manufacturing method thereof

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105334575A (en) * 2015-12-14 2016-02-17 华中科技大学 Silicon-based optical beam splitter and manufacturing method thereof
CN106405730A (en) * 2016-11-28 2017-02-15 华中科技大学 Silicon-based coarse wavelength division device
CN110221384A (en) * 2019-06-17 2019-09-10 华中科技大学 A kind of silicon substrate Meta Materials multimode curved waveguide and preparation method thereof
CN110646883A (en) * 2019-09-30 2020-01-03 华东师范大学重庆研究院 Three-way silicon-based beam splitter chip and manufacturing method thereof

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
DEVICESADRIANO DA SILVA FERREIRA.ET.AL: "Towards an integrated evolutionary strategy and artificial neuralnetwork computational tool for designing photonic coupler devices", APPLIED SOFT COMPUTING *
JASON C. C. MAK.ET.AL: "Binary particle swarm optimized 2 × 2 power splitters in a standard foundry silicon photonic platform", OPTICS LETTERS, vol. 41, no. 16, pages 2 - 3 *
LULUZI LU.ET.AL: "Inverse-designed single-step-etched colorless 3dB couplers based on RIE-lag-insensitive PhC-like subwavelength structures", OPTICS LETTERS, vol. 41, no. 21 *
YINGHENG TANG.ET.AL: "Generative Deep Learning Model for Inverse Design of Integrated Nanophotonic Devices", LASER PHOTONICS REV, vol. 14, no. 12 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116112088A (en) * 2023-04-13 2023-05-12 中国科学院长春光学精密机械与物理研究所 Adaptive modulation method for multi-input multi-output type photon device

Also Published As

Publication number Publication date
CN114325931B (en) 2023-11-17

Similar Documents

Publication Publication Date Title
CN110073359B (en) Efficient data placement for convolutional neural networks
Deng et al. Compressing explicit voxel grid representations: fast nerfs become also small
CN108510064A (en) The processing system and method for artificial neural network including multiple cores processing module
CN109214504A (en) A kind of YOLO network forward inference accelerator design method based on FPGA
JP2016197389A (en) Learning system, learning program, and learning method
CN113627605B (en) Optical interference unit configuration method, device and storage medium of photonic neural network
CN114325931A (en) Manufacturing method of silicon optical device, silicon optical device and photonic integrated circuit
US20230274175A1 (en) Quantum operation control layout for a quantum computation
CN112180592B (en) Rapid optimization method for optical reverse design
CN114519403A (en) Optical diagram neural classification network and method based on-chip diffraction neural network
CN111898316A (en) Construction method and application of super-surface structure design model
CN101071504A (en) Picture special-effect processing method and system
US20240184847A1 (en) Optical circuit building method, optical circuit, and optical signal processing method and apparatus
Sun et al. Multimode diffractive optical neural network
Jahannia et al. Low-latency full precision optical convolutional neural network accelerator
CN113592084B (en) Photon neural network on chip based on reverse optimization superstructure convolution kernel
CN201397417Y (en) Optical system for realizing zooming without moving lens group
CN113325650B (en) Optical circuit, optical signal processing method, optical signal processing device and readable storage medium
US20230259780A1 (en) Neural network sparsification apparatus and method and related product
CN114742219A (en) Neural network computing method and photonic neural network chip architecture
CN113138748B (en) Configurable CNN multiplication accumulator supporting 8bit and 16bit data based on FPGA
CN116112088B (en) Adaptive modulation method for multi-input multi-output type photon device
CN221860699U (en) Beam splitter in vertical direction
CN115641494A (en) Neural network image processing system based on micro-ring modulator
Ma Sr et al. Physics-driven topological optimization of diffractive elements for augmented reality

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant