Through hole contact resistance prediction method, electronic equipment and readable storage medium

Abstract

The invention provides a through hole contact resistance prediction method, electronic equipment and a readable storage medium, and the method comprises the steps: constructing a training sample set based on the obtained historical through hole contact resistance measurement data and historical back-end process online measurement data of a plurality of wafers; training a deep neural network model comprising a feature enhancement layer by adopting a training sample set to obtain a corresponding through hole contact resistance prediction model, the feature enhancement layer is used for learning interaction features among different back-end process parameters based on a multi-head attention mechanism and learning cross-layer relevance between the back-end process parameters and through hole contact resistance; and inputting the current back-end process online measurement data of the to-be-detected wafer into the through hole contact resistance prediction model to obtain through hole contact resistance prediction data of the to-be-detected wafer. According to the invention, real-time accurate prediction of the contact resistance of the through hole can be realized based on deep learning and in combination with an attention mechanism.

Description

Technical Field

[0001] The present invention relates to the technical field of semiconductor processing and manufacturing, and particularly relates to a method for predicting via contact resistance, an electronic device, and a readable storage medium. Background Art

[0002] The via (VIA) contact resistance (Rc) directly affects the electrical performance of the device. Testing and monitoring the via contact resistance is one of the important means to adjust the chip performance and reflect the device process.

[0003] However, there is a significant low-correlation phenomenon between the via contact resistance and the BEOL (Back End of Line) inline measurement data, and there is a lack of interpretable sensitivity correlation. This not only makes it difficult to effectively predict the via contact resistance, but also unable to establish an effective process adjustment strategy when the via contact resistance shows abnormal fluctuations. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for predicting via contact resistance, an electronic device, and a readable storage medium. By combining deep learning with an attention mechanism, it is possible to learn the interaction features between different back-end process parameters and the correlation between cross-layer processes and via contact resistance, so as to achieve real-time and accurate prediction of via contact resistance.

[0005] To achieve the above object, the present invention provides a method for predicting via contact resistance, including: obtaining historical via contact resistance measurement data and historical back-end process inline measurement data of multiple wafers; constructing a training sample set based on the historical via contact resistance measurement data and historical back-end process inline measurement data of the multiple wafers, wherein each training sample in the training sample set includes back-end process inline measurement data and the corresponding via contact resistance measurement data, and the via contact resistance measurement data is the label of the training sample; training a pre-constructed deep neural network model with the training sample set to obtain a corresponding via contact resistance prediction model, wherein the deep neural network model includes a feature enhancement layer, and the feature enhancement layer is used to learn the interaction features between different back-end process parameters and the cross-layer correlation between back-end process parameters and via contact resistance based on the multi-head attention mechanism; inputting the current back-end process inline measurement data of the wafer to be detected into the via contact resistance prediction model to obtain the via contact resistance prediction data of the wafer to be detected.

[0006] Optionally, constructing a training sample set based on the historical through-hole contact resistance measurement data and historical back-end process in-line measurement data of the multiple wafers includes: preprocessing the historical through-hole contact resistance measurement data and historical back-end process in-line measurement data of the multiple wafers; constructing the training sample set according to the preprocessed historical through-hole contact resistance measurement data and historical back-end process in-line measurement data of the multiple wafers.

[0007] Optionally, the preprocessing of the historical through-hole contact resistance measurement data and historical back-end process in-line measurement data of the multiple wafers includes: sequentially performing data cleaning, normalization, and annotation on the historical through-hole contact resistance measurement data and historical back-end process in-line measurement data of the multiple wafers.

[0008] Optionally, the back-end process parameters include the through-hole depth, through-hole diameter, thickness of the bottom metal layer of the through-hole, width of the bottom metal layer of the through-hole, thickness of the top metal layer of the through-hole, and width of the top metal layer of the through-hole.

[0009] Optionally, the through-hole contact resistance prediction method provided by the present invention further includes: obtaining the influence weight information of each back-end process parameter on the through-hole contact resistance according to the attention weight information of the through-hole contact resistance prediction model.

[0010] Optionally, the through-hole contact resistance prediction method provided by the present invention further includes: when the error between the predicted through-hole contact resistance data of the wafer to be detected and the target through-hole contact resistance data of the wafer to be detected exceeds a preset range, formulating a process optimization strategy according to the back-end process parameter with the largest influence weight on the through-hole contact resistance.

[0011] Optionally, the through-hole contact resistance prediction method provided by the present invention further includes: obtaining a cross-layer correlation analysis heat map between the back-end process parameters and the through-hole contact resistance according to the attention weight information of the through-hole contact resistance prediction model; analyzing the cross-layer causality between the through-hole contact resistance and the back-end process parameters according to the cross-layer correlation analysis heat map.

[0012] Optionally, the through-hole contact resistance prediction method provided by the present invention further includes: when the absolute value of the average error between the predicted through-hole contact resistance data of the wafer to be detected and the measured through-hole contact resistance data of the wafer to be detected is greater than a preset threshold, updating the through-hole contact resistance prediction model.

[0013] To achieve the above object, the present invention further provides an electronic device, including a processor and a memory, where a computer program is stored on the memory, and when the computer program is executed by the processor, the through-hole contact resistance prediction method described above is implemented.

[0014] To achieve the above object, the present invention further provides a readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the through-hole contact resistance prediction method described above is implemented.

[0015] Compared with the prior art, the through-hole contact resistance prediction method, electronic device and readable storage medium provided by the present invention have the following unexpected technical effects: The through-hole contact resistance prediction method provided by the present invention first obtains the historical through-hole contact resistance measurement data and historical post-process online measurement data of multiple wafers, and then constructs a training sample set based on the historical through-hole contact resistance measurement data and historical post-process online measurement data of the multiple wafers. Then, the training sample set is used to train a pre-constructed deep neural network model to obtain a corresponding through-hole contact resistance prediction model. Among them, the deep neural network model includes a feature enhancement layer, and the feature enhancement layer is used to learn the interaction features between different post-process parameters based on the multi-head attention mechanism and learn the cross-layer correlation between the post-process parameters and the through-hole contact resistance. Finally, the current post-process online measurement data of the wafer to be detected is input into the through-hole contact resistance prediction model, and the through-hole contact resistance prediction data of the wafer to be detected can be obtained. Thus, based on deep learning and combined with the attention mechanism, the present invention can not only learn the interaction features between different post-process parameters but also learn the correlation between the cross-layer process and the through-hole contact resistance, so as to realize the real-time and accurate prediction of the through-hole contact resistance, effectively save the test time and cost, and at the same time can also lay a certain foundation for subsequent process optimization.

[0016] Since the electronic device and readable storage medium provided by the present invention belong to the same inventive concept as the through-hole contact resistance prediction method provided by the present invention, the electronic device and readable storage medium provided by the present invention at least have all the technical effects of the through-hole contact resistance prediction method provided by the present invention. Therefore, the relevant content about the technical effects of the electronic device and readable storage medium provided by the present invention can refer to the relevant description of the technical effects of the through-hole contact resistance prediction method provided by the present invention in the above text, and will not be repeated here. Description of the Drawings

[0017] Figure 1 It is a flowchart of the through-hole contact resistance prediction method provided by an embodiment of the present invention.

[0018] Figure 2 It is a structural block diagram of the deep neural network model provided by an embodiment of the present invention.

[0019] Figure 3 It is a graph of the influence weights of each post-process parameter on the through-hole contact resistance provided by an embodiment of the present invention.

[0020] Figure 4 The thermal map of the cross-layer correlation analysis between the back-end process parameters and the via contact resistance provided by an embodiment of the present invention.

[0021] Figure 5 The comparison chart of the via contact resistance value predicted by the via contact resistance prediction method provided by the present invention and the actually measured via contact resistance value.

[0022] Figure 6 The schematic block diagram of the electronic device provided by an embodiment of the present invention.

[0023] Among them, the reference numerals are explained as follows: input layer - 110; feature enhancement layer - 120; hidden layer - 130; output layer - 140; processor - 210; communication interface - 220; memory - 230; communication bus - 240. Detailed implementation manners

[0024] The following further details the via contact resistance prediction method, electronic device and readable storage medium proposed by the present invention in conjunction with the accompanying drawings and specific implementation manners. According to the following description, the advantages and features of the present invention will be clearer. It should be noted that the structures, ratios, sizes, etc. shown in the drawings of this specification are only used to cooperate with the content disclosed in the specification for those skilled in this technology to understand and read, and are not used to limit the limiting conditions for the implementation of the present invention. Any modification of the structure, change of the proportional relationship or adjustment of the size, in the case of being the same or similar to the effects that the present invention can produce and the purposes that can be achieved, should still fall within the scope covered by the technical content disclosed by the present invention.

[0025] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising a..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the element. The singular forms "a", "an" and "the" include plural objects, the term "or" is generally used in the sense of including "and / or", the term "several" is generally used in the sense of including "at least one", the term "at least two" is generally used in the sense of including "two or more", and in addition, the terms "first", "second", "third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features.

[0026] In addition, in the description of this specification, descriptions with reference to terms such as "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic descriptions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0027] The core idea of the present invention is to provide a through-hole contact resistance prediction method, an electronic device and a readable storage medium, which can learn the correlation between the cross-layer process and the through-hole contact resistance based on deep learning and combined with the attention mechanism, so as to realize the real-time prediction of the through-hole contact resistance.

[0028] It should be noted that the through-hole contact resistance prediction method provided by the present invention can be applied to the electronic device provided by the present invention, wherein the electronic device can be a personal computer, a mobile terminal, etc., and the mobile terminal can be a hardware device such as a mobile phone, a tablet computer, etc. with various operating systems.

[0029] To implement the above idea, the present invention provides a method for predicting via contact resistance. Please refer to Figure 1 , which is a flowchart of the method for predicting via contact resistance provided by an embodiment of the present invention. As Figure 1 shown, the method for predicting via contact resistance provided by the present invention includes the following steps: Step S100, obtaining historical via contact resistance measurement data and historical back-end process in-line measurement data of multiple wafers; Step S200, constructing a training sample set based on the historical via contact resistance measurement data and historical back-end process in-line measurement data of the multiple wafers, wherein each training sample in the training sample set includes back-end process in-line measurement data and corresponding via contact resistance measurement data, and the via contact resistance measurement data is the label of the training sample; Step S300, training a pre-constructed deep neural network model with the training sample set to obtain a corresponding via contact resistance prediction model; Step S400, inputting the current back-end process in-line measurement data of the wafer to be detected into the via contact resistance prediction model to obtain the via contact resistance prediction data of the wafer to be detected.

[0030] Please continue to refer to Figure 2 , which is a structural block diagram of the deep neural network model provided by an embodiment of the present invention. As Figure 2 shown, the deep neural network model includes a feature enhancement layer 120, and the feature enhancement layer 120 is used to learn the interaction features between different back-end process parameters and the cross-layer correlation between the back-end process parameters and the via contact resistance based on the multi-head attention mechanism. Thus, the present invention can not only learn the interaction features between different back-end process parameters but also learn the correlation between the cross-layer process and the via contact resistance based on deep learning and combined with the attention mechanism, so as to realize the real-time and accurate prediction of the via contact resistance, effectively save the test time and cost, and at the same time can also lay a certain foundation for the subsequent process optimization.

[0031] Please continue to refer to Figure 2 , as Figure 2As shown, the deep neural network model further includes an input layer 110, a hidden layer 130, and an output layer 140. The feature enhancement layer 120 is located between the input layer 110 and the hidden layer 130. The input layer 110 is used to receive input data (online measurement data of the back-end process), the hidden layer 130 is used to implement neuron transmission of parameters using the sigmoid activation function, and the output layer 140 is used to output via contact resistance prediction data using the linear activation function. It should be noted that, as can be understood by those skilled in the art, the present invention does not limit the number of hidden layers 130 included in the deep neural network model and the number of neurons included in the hidden layer 130. The number of the hidden layer 130 and the neurons can be adjusted according to the data volume and data complexity.

[0032] Further, the online measurement data of the back-end process in each training sample includes the online measurement data of multiple back-end process parameters corresponding to each layer of vias, and the via contact resistance measurement data in each training sample includes the contact resistance corresponding to each layer of vias.

[0033] Specifically, the online measurement data of N back-end process parameters corresponding to each of the M layers of vias can be converted into a data matrix X after preprocessing:

[0034] where X 11 ,..., X 1N represents the online measurement data of N back-end process parameters corresponding to the first layer of vias; X M1 ,..., X MN represents the online measurement data of N back-end process parameters corresponding to the Mth layer of vias.

[0035] Taking the 2-head attention mechanism as an example, the dimension of each head is taken as N / 2, focusing on the back-end process parameters of the current layer and the cross-layer back-end process parameters respectively.

[0036] The weight matrix W 1-Q of the query vector of the first head, the weight matrix W 1-K of the key vector, and the weight matrix W 1-V of the value vector are as follows:

[0037]

[0038]

[0039] Then the query vector Q1 of the first head = X × W 1-Q , and the key vector K1 of the first head = X × W 1-K, the value vector V1 of the first head = X × W 1-V .

[0040] The calculation formula for the attention head1 of the first head is as follows:

[0041] Similarly, the attention head2 of the second head can be calculated based on the following formula:

[0042] Among them, the query vector Q2 of the second head = X × W 2-Q , W 2-Q is the weight matrix of the query vector of the second head; the key vector K2 of the second head = X × W 2-K , W 2-K is the weight matrix of the key vector of the second head; the value vector V2 of the second head = X × W 2-V , W 2-V is the weight matrix of the value vector of the second head.

[0043] The final output result FO of the feature enhancement layer 120 = Concat(head1, head2). It should be noted that the dimension of the concatenation (Concat) result of the attention head1 of the first head and the attention head2 of the second head is still N.

[0044] The hidden layer 130 performs a non-linear transformation on the output result FO of the feature enhancement layer 120 according to the following formula:

[0045] Among them, Y is the output of the hidden layer 130, W M*N is the weight matrix learned by the hidden layer 130, and b is the bias term.

[0046] Thus, through the cyclic calculation of multiple hidden layers 130 and the linear calculation of the output layer 140, the through-hole contact resistance prediction data can be output.

[0047] In some exemplary embodiments, the post-process parameters include the through-hole depth, through-hole diameter, thickness of the bottom metal layer of the through-hole, width of the bottom metal layer of the through-hole, thickness of the top metal layer of the through-hole, and width of the top metal layer of the through-hole.

[0048] It should be noted that, as can be understood by those skilled in the art, the post-process parameters may include other parameters in addition to the through-hole depth, through-hole diameter, thickness of the bottom metal layer of the through-hole, width of the bottom metal layer of the through-hole, thickness of the top metal layer of the through-hole, and width of the top metal layer of the through-hole, such as yellow light parameters, etching parameters, deposition parameters, material property parameters, etc.

[0049] In some exemplary embodiments, constructing a training sample set based on the historical through-hole contact resistance measurement data and historical back-end process in-line measurement data of the multiple wafers includes: preprocessing the historical through-hole contact resistance measurement data and historical back-end process in-line measurement data of the multiple wafers; and constructing the training sample set according to the preprocessed historical through-hole contact resistance measurement data and historical back-end process in-line measurement data of the multiple wafers.

[0050] Thus, by first preprocessing the historical through-hole contact resistance measurement data and historical back-end process in-line measurement data of the multiple wafers, and then constructing the training sample set according to the preprocessed historical through-hole contact resistance measurement data and historical back-end process in-line measurement data of the multiple wafers, a high-quality training sample set can be provided for subsequent model training, thereby improving the prediction accuracy of the through-hole contact resistance prediction model obtained by training.

[0051] In some exemplary embodiments, preprocessing the historical through-hole contact resistance measurement data and historical back-end process in-line measurement data of the multiple wafers includes: sequentially performing data cleaning, normalization, and annotation on the historical through-hole contact resistance measurement data and historical back-end process in-line measurement data of the multiple wafers.

[0052] Thus, through data cleaning, noise and outliers introduced due to equipment fluctuations, measurement errors, or human operation mistakes can be removed, preventing the model from being misled by incorrect data during subsequent training, and thus reducing the prediction deviation of the model. By performing a normalization operation on the data, different types of data can be mapped to the same numerical range (e.g., [0, 1]), thereby accelerating the convergence speed of the model during subsequent training. By annotating the data, the in-line measurement data of the back-end process can be accurately matched with the corresponding through-hole contact resistance, ensuring that the input features and labels of each training sample correspond one by one. It should be noted that, as can be understood by those skilled in the art, the data needs to be aligned according to the process timing during annotation (e.g., photolithography first, etching later) to ensure that the model can capture the dynamic effects of process steps.

[0053] Further, before inputting the current in-line measurement data of the back-end process of the wafer to be detected into the through-hole contact resistance prediction model, the through-hole contact resistance prediction method further includes: normalizing the current in-line measurement data of the back-end process of the wafer to be detected.

[0054] Correspondingly, inputting the current in-line measurement data of the back-end process of the wafer to be detected into the through-hole contact resistance prediction model includes: inputting the normalized current in-line measurement data of the back-end process of the wafer to be detected into the through-hole contact resistance prediction model.

[0055] In some exemplary embodiments, the method for predicting via contact resistance provided by the present invention further includes: obtaining the influence weight information of each back-end process parameter on the via contact resistance according to the attention weight information of the via contact resistance prediction model.

[0056] Thus, by obtaining the influence weight information of each back-end process parameter on the via contact resistance according to the attention weight information of the via contact resistance prediction model, engineers can intuitively determine which back-end process parameters have a greater impact on the via contact resistance, thereby providing a theoretical basis for subsequent process optimization.

[0057] Specifically, please refer to Figure 3 , which is a graph showing the influence weights of each back-end process parameter on the via contact resistance provided by an embodiment of the present invention. As Figure 3 shown, the back-end process parameter A has the greatest influence weight on the via contact resistance, the back-end process parameter B has the second greatest influence weight on the via contact resistance, and the influence weights of the back-end process parameters C, D, E, and F on the via contact resistance decrease in sequence. Thus, according to the influence weights of each back-end process parameter on the via contact resistance, the back-end process parameters with relatively large influence weights on the via contact resistance can be selected to determine the process optimization strategy. It should be noted that Figure 3 the influence weight values of each back-end process parameter on the via contact resistance in

[0058] In some exemplary embodiments, the method for predicting via contact resistance provided by the present invention further includes: when the error between the predicted via contact resistance data of the wafer to be detected and the target via contact resistance data of the wafer to be detected exceeds a preset range, formulating a process optimization strategy according to the back-end process parameter with the greatest influence weight on the via contact resistance.

[0059] Thus, by formulating a process optimization strategy according to the back-end process parameter with the greatest influence weight on the via contact resistance when the error between the predicted via contact resistance data of the wafer to be detected and the target via contact resistance data of the wafer to be detected exceeds a preset range, not only can the process optimization time be effectively saved, but also the effectiveness of the process optimization can be effectively ensured.

[0060] In some exemplary embodiments, the method for predicting via contact resistance provided by the present invention further includes: obtaining a heat map of cross-layer correlation analysis between the back-end process parameter and the via contact resistance according to the attention weight information of the via contact resistance prediction model; analyzing the cross-layer causality between the via contact resistance and the back-end process parameter according to the heat map of cross-layer correlation analysis.

[0061] Therefore, by analyzing the cross-layer causality between the via contact resistance and the back-end process parameters, direct causality and indirect correlation can be distinguished, so that engineers can be guided to preferentially monitor the back-end process parameters that have a direct causal relationship with the via contact resistance, thereby providing a clear direction for process optimization.

[0062] Please continue to refer to Figure 4 , which is a heat map of the cross-layer correlation analysis between the back-end process parameters and the via contact resistance provided by an embodiment of the present invention. As Figure 4 shown, through correlation analysis, it is found that there is a correlation (0.48) between Via3_thk (the thickness of the via in the third layer) and Rc_V2 (the contact resistance of the via in the second layer), and there is a strong correlation (0.72) between Via3_thk (the thickness of the via in the third layer) and Via2_thk (the thickness of the via in the second layer). Further evaluation reveals that there is no causality between Via3_thk (the thickness of the via in the third layer) and Rc_V2 (the contact resistance of the via in the second layer), and the correlation between Via3_thk (the thickness of the via in the third layer) and Rc_V2 (the contact resistance of the via in the second layer) is due to the similar process types of Via3_thk (the thickness of the via in the third layer) and Via2_thk (the thickness of the via in the second layer). It should be noted that, as can be understood by those skilled in the art, Figure 4 the various numerical values shown are only for illustrative purposes and do not constitute a limitation to the present invention.

[0063] In some exemplary embodiments, the method for predicting the via contact resistance provided by the present invention further includes: when the absolute value of the average error between the predicted data of the via contact resistance of the wafer to be detected and the measured data of the via contact resistance of the wafer to be detected is greater than a preset threshold, updating the via contact resistance prediction model.

[0064] Therefore, when the absolute value of the average error between the predicted data of the via contact resistance of the wafer to be detected and the measured data of the via contact resistance of the wafer to be detected is greater than a preset threshold (e.g., 5%), updating the via contact resistance prediction model can improve the robustness of the via contact resistance prediction model, enabling the via contact resistance prediction model to maintain high prediction accuracy and avoid prediction deviation.

[0065] Specifically, the via contact resistance prediction model can be updated by strategies such as hyperparameter tuning, model architecture optimization (such as attention mechanism adjustment, hidden layer 130 structure adjustment), data augmentation and regularization, and training strategy optimization (such as optimizer selection, learning rate adjustment, batch adjustment).

[0066] Please continue to refer to Table 1, which shows the comparison results between the via contact resistance values predicted by the via contact resistance prediction method provided by the present invention (abbreviated as predicted values in Table 1) and the actual measured via contact resistance values (abbreviated as actual measured values in Table 1). As shown in Table 1, the accuracy rate of the via contact resistance data predicted by the via contact resistance prediction method provided by the present invention is greater than 93%, meeting the controllability requirements and being able to improve the volatility of the via contact resistance.

[0067] Table 1 Comparison Results between Predicted Values and Actual Measured Values of Via Contact Resistance

[0068] Please continue to refer to Figure 5 , which is a comparison chart of the via contact resistance values predicted by the via contact resistance prediction method provided by the present invention and the actual measured via contact resistance values. As Figure 5 shown, the via contact resistance data predicted by the via contact resistance prediction method provided by the present invention is highly consistent with the actual measured via contact resistance data.

[0069] Based on the same inventive concept, the present invention also provides an electronic device. Please refer to Figure 6 , which is a schematic block diagram of the electronic device provided by an embodiment of the present invention. As Figure 6 shown, the electronic device includes a processor 210 and a memory 230. A computer program is stored on the memory 230. When the computer program is executed by the processor 210, it implements the above-mentioned via contact resistance prediction method. Since the electronic device provided by the present invention and the via contact resistance prediction method provided by the present invention belong to the same inventive concept, the electronic device provided by the present invention has at least all the beneficial effects of the via contact resistance prediction method provided by the present invention. Therefore, for the beneficial effects of the electronic device provided by the present invention, reference can be made to the relevant descriptions of the beneficial effects of the via contact resistance prediction method provided by the present invention above, and no further elaboration will be provided here.

[0070] Please continue to refer to Figure 6 , as Figure 6As shown, the electronic device further includes a communication interface 220 and a communication bus 240. Among them, the processor 210, the communication interface 220, and the memory 230 complete mutual communication through the communication bus 240. The communication bus 240 may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus 240 can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, only a thick line is used in the figure, but it does not mean that there is only one bus or one type of bus. The communication interface 220 is used for communication between the above-mentioned electronic device and other devices.

[0071] It should be noted that the processor 210 referred to in the present invention may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor 210 is the control center of the electronic device, and uses various interfaces and lines to connect all parts of the entire electronic device.

[0072] It should also be noted that the memory 230 can be used to store the computer program. By running or executing the computer program stored in the memory 230 and calling the data stored in the memory 230, the processor 210 realizes various functions of the electronic device. The memory 230 may include non-volatile and / or volatile memory. The non-volatile memory may include read-only memory (ROM), programmable memory (PROM), electrically programmable memory (EPROM), electrically erasable programmable memory (EEPROM), or flash memory. The volatile memory may include random access memory (RAM) or an external cache memory. By way of illustration and not limitation, random access memory is available in various forms, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous random access memory (SDRAM), double data rate synchronous random access memory (DDR SDRAM), enhanced synchronous random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), Rambus direct random access memory (RDRAM), direct memory bus dynamic random access memory (DRDRAM), and Rambus dynamic random access memory (RDRAM), etc.

[0073] The present invention also provides a readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, it can implement the through-hole contact resistance prediction method described above. Since the readable storage medium provided by the present invention and the through-hole contact resistance prediction method provided by the present invention belong to the same inventive concept, the readable storage medium provided by the present invention has at least all the beneficial effects of the through-hole contact resistance prediction method provided by the present invention. Therefore, for the beneficial effects of the readable storage medium provided by the present invention, reference can be made to the relevant descriptions of the beneficial effects of the through-hole contact resistance prediction method provided by the present invention in the above text, and no further elaboration will be made here.

[0074] The readable storage medium provided by the present invention may adopt any combination of one or more computer-readable media. The readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may, for example, but not be limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the computer-readable storage medium include: an electrical connection having one or more wires, a portable computer hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this article, the computer-readable storage medium may be any tangible medium that contains or stores a program, and this program can be used by or combined with an instruction execution system, apparatus, or device.

[0075] The computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer-readable signal medium may also be any computer-readable medium other than the computer-readable storage medium, and this computer-readable medium can send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted by any appropriate medium, including but not limited to wireless, wire, optical cable, etc., or any suitable combination of the above.

[0076] In summary, compared with the prior art, the through-hole contact resistance prediction method, electronic device, and readable storage medium provided by the present invention have the following unexpected technical effects: By combining deep learning with an attention mechanism, the present invention can not only learn the interaction characteristics between different back-end process parameters but also learn the correlation between the cross-layer process and the through-hole contact resistance, thereby enabling real-time and accurate prediction of the through-hole contact resistance, effectively saving test time and cost, and at the same time laying a certain foundation for subsequent process optimization.

[0077] It should be noted that computer program code for performing the operations of the present invention can be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, by connecting through an Internet service provider via the Internet).

[0078] It should be noted that the above description is only a description of the preferred embodiments of the present invention and does not limit the scope of the present invention in any way. Any changes and modifications made by those of ordinary skill in the art of the present invention based on the above disclosure fall within the protection scope of the present invention. Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations are within the scope of the present invention and its equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A method for predicting the through-hole contact resistance, characterized in that Including: Obtaining historical through - hole contact resistance measurement data and historical back - end process in - line measurement data of multiple wafers; Based on the historical through - hole contact resistance measurement data and historical back - end process in - line measurement data of the multiple wafers, constructing a training sample set, wherein each training sample in the training sample set includes back - end process in - line measurement data and corresponding through - hole contact resistance measurement data, and the through - hole contact resistance measurement data is the label of the training sample; Using the training sample set to train a pre - constructed deep neural network model to obtain a corresponding through - hole contact resistance prediction model, wherein the deep neural network model includes a feature enhancement layer, and the feature enhancement layer is used to learn the interaction features between different back - end process parameters and the cross - layer correlation between the back - end process parameters and the through - hole contact resistance based on the multi - head attention mechanism; Inputting the current back - end process in - line measurement data of the wafer to be detected into the through - hole contact resistance prediction model to obtain the through - hole contact resistance prediction data of the wafer to be detected.

2. The through-hole contact resistance prediction method according to claim 1, wherein The constructing a training sample set based on the historical through - hole contact resistance measurement data and historical back - end process in - line measurement data of the multiple wafers includes: Pre - processing the historical through - hole contact resistance measurement data and historical back - end process in - line measurement data of the multiple wafers; Constructing the training sample set according to the pre - processed historical through - hole contact resistance measurement data and historical back - end process in - line measurement data of the multiple wafers.

3. The through-hole contact resistance prediction method according to claim 2, characterized in that The pre - processing the historical through - hole contact resistance measurement data and historical back - end process in - line measurement data of the multiple wafers includes: Successively performing data cleaning, normalization, and annotation on the historical through - hole contact resistance measurement data and historical back - end process in - line measurement data of the multiple wafers.

4. The through-hole contact resistance prediction method according to claim 1, wherein The back - end process parameters include through - hole depth, through - hole diameter, thickness of the bottom metal layer of the through - hole, width of the bottom metal layer of the through - hole, thickness of the top metal layer of the through - hole, and width of the top metal layer of the through - hole.

5. The through-hole contact resistance prediction method according to claim 1, wherein The method further includes: Obtaining the influence weight information of each back - end process parameter on the through - hole contact resistance according to the attention weight information of the through - hole contact resistance prediction model.

6. The through-hole contact resistance prediction method according to claim 5, characterized in that The method further includes: When the error between the through - hole contact resistance prediction data of the wafer to be detected and the target data of the through - hole contact resistance of the wafer to be detected exceeds a preset range, formulating a process optimization strategy according to the back - end process parameter with the largest influence weight on the through - hole contact resistance.

7. The through-hole contact resistance prediction method according to claim 5, characterized in that The method further includes: Obtaining a cross - layer correlation analysis heat map between the back - end process parameters and the through - hole contact resistance according to the attention weight information of the through - hole contact resistance prediction model; Analyzing the cross - layer causality between the through - hole contact resistance and the back - end process parameters according to the cross - layer correlation analysis heat map.

8. The method for predicting the through-hole contact resistance according to claim 1, wherein The method further includes: When the absolute value of the average error between the through - hole contact resistance prediction data of the wafer to be detected and the measured data of the through - hole contact resistance of the wafer to be detected is greater than a preset threshold, updating the through - hole contact resistance prediction model.

9. An electronic device, characterized in that, It includes a processor and a memory, and a computer program is stored on the memory. When the computer program is executed by the processor, it implements the through-hole contact resistance prediction method according to any one of claims 1 to 8.

10. A readable storage medium, characterized in that, A computer program is stored in the readable storage medium. When the computer program is executed by a processor, it implements the through-hole contact resistance prediction method according to any one of claims 1 to 8.

Images