CN118230855A - Wafer cleaning method for silicon part - Google Patents
Wafer cleaning method for silicon part Download PDFInfo
- Publication number
- CN118230855A CN118230855A CN202410351527.2A CN202410351527A CN118230855A CN 118230855 A CN118230855 A CN 118230855A CN 202410351527 A CN202410351527 A CN 202410351527A CN 118230855 A CN118230855 A CN 118230855A
- Authority
- CN
- China
- Prior art keywords
- silicon nitride
- cleaning
- nitride layer
- wafer
- training data
- 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.)
- Pending
Links
- 238000004140 cleaning Methods 0.000 title claims abstract description 164
- 238000000034 method Methods 0.000 title claims abstract description 76
- XUIMIQQOPSSXEZ-UHFFFAOYSA-N Silicon Chemical group [Si] XUIMIQQOPSSXEZ-UHFFFAOYSA-N 0.000 title claims abstract description 17
- 229910052581 Si3N4 Inorganic materials 0.000 claims abstract description 148
- HQVNEWCFYHHQES-UHFFFAOYSA-N silicon nitride Chemical compound N12[Si]34N5[Si]62N3[Si]51N64 HQVNEWCFYHHQES-UHFFFAOYSA-N 0.000 claims abstract description 148
- 238000012549 training Methods 0.000 claims abstract description 127
- 238000012545 processing Methods 0.000 claims abstract description 100
- 238000012216 screening Methods 0.000 claims abstract description 14
- 235000012431 wafers Nutrition 0.000 claims description 126
- 239000011159 matrix material Substances 0.000 claims description 44
- 238000013441 quality evaluation Methods 0.000 claims description 32
- 238000004364 calculation method Methods 0.000 claims description 22
- 238000000137 annealing Methods 0.000 claims description 16
- 238000005137 deposition process Methods 0.000 claims description 15
- 238000011156 evaluation Methods 0.000 claims description 12
- 238000010276 construction Methods 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 abstract description 5
- 238000005457 optimization Methods 0.000 description 3
- VYPSYNLAJGMNEJ-UHFFFAOYSA-N Silicium dioxide Chemical compound O=[Si]=O VYPSYNLAJGMNEJ-UHFFFAOYSA-N 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 229910052814 silicon oxide Inorganic materials 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 229910052710 silicon Inorganic materials 0.000 description 1
- 239000010703 silicon Substances 0.000 description 1
Landscapes
- Testing Or Measuring Of Semiconductors Or The Like (AREA)
Abstract
The invention discloses a wafer cleaning method of a silicon part, which relates to the field of wafer cleaning and comprises the following steps: constructing a parameter training database; acquiring processing parameters of the silicon nitride layer of the wafer, and recording the processing parameters as real-time processing parameters of the silicon nitride layer of the wafer; according to the real-time processing parameters of the silicon nitride layer of the wafer, a plurality of parameter training data sets are called from a parameter training database; screening a plurality of optimized training data from the parameter training data set; calculating according to the optimized training data to obtain cleaning parameters of the silicon nitride layer; performing a silicon nitride layer cleaning process according to the silicon nitride layer cleaning parameters; updating the parameter training database, and calculating the cleaning parameters of the silicon nitride layer by adopting the updated parameter training database. The invention has the advantages that: the intelligent analysis is carried out by combining the processing data of the wafer cleaning process, and the optimized processing parameter values of the wafer cleaning process can be automatically set after the silicon nitride layer is processed, so that the cleaning quality of the wafer can be effectively ensured, and the subsequent processing yield of the wafer is improved.
Description
Technical Field
The invention relates to the field of wafer cleaning, in particular to a wafer cleaning method for silicon parts.
Background
In practical applications, a silicon nitride layer with tensile stress is generally deposited on the upper surface of a wafer, then the stress in the silicon nitride layer is kept in the wafer through an annealing process, and finally the silicon nitride layer is removed through a cleaning process.
Since the silicon nitride layer is generally prepared by a furnace process, both the upper and lower surfaces of the wafer will form the silicon nitride layer. When the silicon nitride layer on the upper surface of the wafer is removed by the cleaning process, the upper surface and the lower surface of the wafer are soaked in the cleaning liquid at the same time for cleaning. However, the silicon nitride layer on the lower surface of the wafer cannot be completely removed due to the limited cleaning process, so that the silicon nitride layer remains more obviously, especially at the edge of the wafer, and the remaining silicon nitride layer remains until the subsequent process, which affects the yield of semiconductor devices.
In practical application, different deposition processes and annealing processes can lead to silicon nitride layers with different performances, in the prior art, intelligent analysis on a wafer cleaning process is lacking, process parameters of the wafer cleaning process are difficult to adjust in real time and intelligently according to wafer cleaning data, the maximum effect of wafer cleaning is difficult to achieve, and the processing yield of wafers is affected.
Disclosure of Invention
In order to solve the technical problems, the wafer cleaning method for the silicon component is provided, and the technical scheme solves the problems that in the prior art, intelligent analysis on a wafer cleaning process is lacking, real-time intelligent learning adjustment of process parameters of the wafer cleaning process is difficult to be performed according to wafer cleaning data, the maximum effect of wafer cleaning is difficult to be achieved, and the processing yield of wafers is affected.
In order to achieve the above purpose, the invention adopts the following technical scheme:
A wafer cleaning method of a silicon part, comprising:
Constructing a parameter training database, wherein the parameter training database is used for storing historical processing parameters and historical cleaning data of a silicon nitride layer of a wafer;
acquiring processing parameters of a silicon nitride layer of a wafer, and recording the processing parameters as real-time processing parameters of the silicon nitride layer of the wafer, wherein the processing parameters of the silicon nitride layer of the wafer at least comprise deposition process parameters and annealing process parameters;
According to the real-time processing parameters of the silicon nitride layer of the wafer, a plurality of parameter training data sets are called from a parameter training database;
screening a plurality of optimized training data from the parameter training data set;
calculating according to the optimized training data to obtain cleaning parameters of the silicon nitride layer;
performing a silicon nitride layer cleaning process according to the silicon nitride layer cleaning parameters;
acquiring cleaning processing data under the current silicon nitride layer cleaning process, and adding the cleaning processing data under the current silicon nitride layer cleaning process into a parameter training database to obtain an updated parameter training database;
and when new cleaning of the silicon nitride layer is carried out, calculating cleaning parameters of the silicon nitride layer by adopting the updated parameter training database.
Preferably, the retrieving, from the parameter training database, a plurality of parameter training data sets corresponding to the current silicon nitride layer according to the current processing parameters of the silicon nitride layer of the wafer specifically includes:
Calculating the similarity between the real-time processing parameters of the silicon nitride layers of the wafers and the historical processing parameters of the silicon nitride layers of each wafer respectively through a processing similarity calculation formula;
Screening a plurality of historical processing parameters of the wafer silicon nitride layer with the similarity smaller than a preset value, and marking the historical processing parameters as fitting processing parameters of the wafer silicon nitride layer;
Using historical cleaning data corresponding to the wafer silicon nitride layer fitting processing parameters as a parameter training data set;
preferably, the processing similarity calculation formula specifically includes:
Wherein L a is the similarity between the real-time processing parameters of the silicon nitride layer of the wafer and the historical processing parameters of the silicon nitride layer of the wafer a, x i0 is the i deposition process parameter value in the real-time processing parameters of the silicon nitride layer of the wafer a, x ia is the i deposition process parameter value in the historical processing parameters of the silicon nitride layer of the wafer a, n is the total number of deposition process parameters, y j0 is the j annealing process parameter value in the real-time processing parameters of the silicon nitride layer of the wafer a, y ja is the j annealing process parameter value in the historical processing parameters of the silicon nitride layer of the wafer a, and m is the total number of annealing process parameters.
Preferably, the screening the plurality of optimized training data from the parameter training data set specifically includes:
determining a plurality of quality evaluation indexes of wafer cleaning, wherein the quality evaluation indexes at least comprise residual quantity of a silicon nitride layer and time consumption of wafer cleaning;
Extracting the residual quantity of the silicon nitride layer and the time consumption for cleaning the wafer from the parameter training data set to form a quality evaluation matrix A;
Carrying out standardization processing on the quality evaluation matrix A to obtain a standardization evaluation matrix B;
Based on the standardized evaluation matrix B, calculating a comprehensive cleaning quality index corresponding to each parameter training data set;
Setting an optimized training total number N, screening out parameter training data sets with the top N bits of comprehensive cleaning quality indexes according to the sequence from small to large, and taking the parameter training data sets as optimized training data.
Preferably, the normalizing the quality evaluation matrix a to obtain a normalized evaluation matrix B specifically includes:
the quality evaluation matrix Wherein a l1 is the residual quantity of the silicon nitride layer of the 1 st parameter training data set clock, a l2 is the time consumption of wafer cleaning of the 1 st parameter training data set clock, and k is the total number of parameter training data sets;
Respectively calculating a silicon nitride layer residue standardized value and a wafer cleaning time-consuming standardized value through a standardized formula;
The standardized formula is:
Wherein b l1 is a standardized value of the residual quantity of the silicon nitride layer in the 1 st parameter training data set, and b l2 is a standardized value of the time consumption of wafer cleaning in the 1 st parameter training data set;
A quality evaluation matrix B is constructed and is used for the construction of a quality evaluation matrix,
Preferably, calculating the comprehensive cleaning quality index corresponding to each parameter training data set based on the standardized evaluation matrix B specifically includes:
Based on the quality evaluation matrix B, an idealized optimal quality matrix M + is screened out;
Based on a distance calculation formula, respectively calculating a vector distance between the silicon nitride layer residue standardized value and the wafer cleaning time-consuming standardized value of each parameter training data set and an idealized optimal quality matrix M +, and taking the vector distance as a comprehensive cleaning quality index corresponding to the parameter training data set.
Preferably, the distance calculation formula specifically includes:
Wherein Z l is the comprehensive cleaning quality index of the 1 st parameter training data set.
Preferably, the calculating according to the optimized training data to obtain the cleaning parameters of the silicon nitride layer specifically includes:
Extracting historical parameter values corresponding to each silicon nitride layer cleaning parameter in the optimized training data to serve as silicon nitride layer cleaning parameter training data;
Calculating cleaning parameters of the silicon nitride layer through a parameter calculation formula based on the training data of the cleaning parameters of the silicon nitride layer;
the parameter calculation formula is as follows:
Wherein, W g is the parameter value of the cleaning parameter of the g-th silicon nitride layer, and W gu is the history parameter value corresponding to the cleaning parameter of the g-th silicon nitride layer in the u-th optimized training data.
Compared with the prior art, the invention has the beneficial effects that:
According to the wafer cleaning method for the silicon part, real-time processing parameters of the silicon nitride layer of the wafer are matched with historical processing parameters of the silicon nitride layer of the wafer, parameter values of wafer cleaning processes corresponding to the historical processing states closest to the processing states of the silicon nitride layer of the current wafer are screened out to serve as parameter training data sets, parameter values of a plurality of groups of wafer cleaning processes with highest cleaning quality are screened out from the parameter training data sets through a distance optimization algorithm to serve as optimized training data, an average value is calculated according to the optimized training data to serve as cleaning process parameters of the wafer, intelligent analysis is conducted on the wafer cleaning process processing data, optimized wafer cleaning process processing parameter values can be automatically set after the processing of the silicon nitride layer is finished, the cleaning quality of the wafer can be effectively guaranteed, and the subsequent processing yield of the wafer is improved.
Drawings
FIG. 1 is a flow chart of a method for cleaning a silicon part wafer according to the present invention;
FIG. 2 is a flow chart of a method for retrieving a plurality of parameter training data sets according to the present invention;
FIG. 3 is a flow chart of a method for screening out a plurality of optimized training data from a parameter training data set according to the present invention;
FIG. 4 is a flowchart of a method for calculating a comprehensive cleaning quality index corresponding to each parameter training data set according to the present invention;
FIG. 5 is a flowchart of a method for obtaining cleaning parameters of a silicon nitride layer according to the calculation of optimized training data in the present invention.
Detailed Description
The following description is presented to enable one of ordinary skill in the art to make and use the invention. The preferred embodiments in the following description are by way of example only and other obvious variations will occur to those skilled in the art.
Referring to fig. 1, a wafer cleaning method for a silicon part includes:
constructing a parameter training database, wherein the parameter training database is used for storing historical processing parameters and historical cleaning data of a silicon nitride layer of a wafer;
acquiring processing parameters of the silicon nitride layer of the wafer, and recording the processing parameters as real-time processing parameters of the silicon nitride layer of the wafer, wherein the processing parameters of the silicon nitride layer of the wafer at least comprise deposition process parameters and annealing process parameters;
According to the real-time processing parameters of the silicon nitride layer of the wafer, a plurality of parameter training data sets are called from a parameter training database;
screening a plurality of optimized training data from the parameter training data set;
calculating according to the optimized training data to obtain cleaning parameters of the silicon nitride layer;
performing a silicon nitride layer cleaning process according to the silicon nitride layer cleaning parameters;
acquiring cleaning processing data under the current silicon nitride layer cleaning process, and adding the cleaning processing data under the current silicon nitride layer cleaning process into a parameter training database to obtain an updated parameter training database;
and when new cleaning of the silicon nitride layer is carried out, calculating cleaning parameters of the silicon nitride layer by adopting the updated parameter training database.
According to the scheme, the real-time processing parameters of the silicon nitride layer of the wafer are matched with the historical processing parameters of the silicon nitride layer of the wafer, the parameter values of the wafer cleaning process corresponding to the historical processing state closest to the current processing state of the silicon nitride layer of the wafer are screened out to serve as parameter training data sets, the parameter values of a plurality of groups of wafer cleaning processes with the highest cleaning quality are screened out from the parameter training data sets to serve as optimized training data through a distance optimization algorithm, the cleaning process parameters of the wafer are calculated according to the optimized training data, and the optimized setting of the processing parameter values of the wafer cleaning process can be achieved.
Referring to fig. 2, the step of retrieving a plurality of parameter training data sets corresponding to the current silicon nitride layer from a parameter training database according to the current wafer silicon nitride layer processing parameters specifically includes:
Calculating the similarity between the real-time processing parameters of the silicon nitride layers of the wafers and the historical processing parameters of the silicon nitride layers of each wafer respectively through a processing similarity calculation formula;
Screening a plurality of historical processing parameters of the wafer silicon nitride layer with the similarity smaller than a preset value, and marking the historical processing parameters as fitting processing parameters of the wafer silicon nitride layer;
Using historical cleaning data corresponding to the wafer silicon nitride layer fitting processing parameters as a parameter training data set;
The processing similarity calculation formula specifically comprises:
Wherein L a is the similarity between the real-time processing parameters of the silicon nitride layer of the wafer and the historical processing parameters of the silicon nitride layer of the wafer a, x i0 is the i deposition process parameter value in the real-time processing parameters of the silicon nitride layer of the wafer a, x ia is the i deposition process parameter value in the historical processing parameters of the silicon nitride layer of the wafer a, n is the total number of deposition process parameters, y j0 is the j annealing process parameter value in the real-time processing parameters of the silicon nitride layer of the wafer a, y ja is the j annealing process parameter value in the historical processing parameters of the silicon nitride layer of the wafer a, and m is the total number of annealing process parameters.
It can be understood that the structural properties of the silicon nitride layer are closely related to the deposition process parameter and the annealing process parameter, the structural properties of the silicon nitride layer with similar deposition process parameter and annealing process parameter are the same, and the cleaning process parameter required during cleaning is the same.
Referring to fig. 3, screening a plurality of optimized training data from a parameter training data set specifically includes:
Determining a plurality of quality evaluation indexes of wafer cleaning, wherein the quality evaluation indexes at least comprise residual quantity of a silicon nitride layer and time consumption of wafer cleaning;
in some embodiments, extracting the residual amount of the silicon nitride layer and the time consumed for cleaning the wafer from the parameter training data set to form a quality evaluation matrix A;
Carrying out standardization processing on the quality evaluation matrix A to obtain a standardization evaluation matrix B;
Based on the standardized evaluation matrix B, calculating a comprehensive cleaning quality index corresponding to each parameter training data set;
Setting an optimized training total number N, screening out parameter training data sets with the top N bits of comprehensive cleaning quality indexes according to the sequence from small to large, and taking the parameter training data sets as optimized training data.
The quality evaluation matrix A is subjected to standardization processing, and the standardized evaluation matrix B is obtained specifically comprises the following steps:
Quality evaluation matrix Wherein a l1 is the residual quantity of the silicon nitride layer of the 1 st parameter training data set clock, a l2 is the time consumption of wafer cleaning of the 1 st parameter training data set clock, and k is the total number of parameter training data sets;
Respectively calculating a silicon nitride layer residue standardized value and a wafer cleaning time-consuming standardized value through a standardized formula;
The normalized formula is:
Wherein b l1 is a standardized value of the residual quantity of the silicon nitride layer in the 1 st parameter training data set, and b l2 is a standardized value of the time consumption of wafer cleaning in the 1 st parameter training data set;
A quality evaluation matrix B is constructed and is used for the construction of a quality evaluation matrix,
It can be understood that, for the wafers with different cleaning requirements, the quality evaluation indexes can be further set to other indexes, such as the residual quantity of the cleaning solution and the damage degree of the wafer, in this scheme, the residual quantity of the silicon nitride layer and the time consumed for cleaning the wafer are taken as examples, the quality evaluation matrix a is constructed, the quality evaluation matrix B is obtained by standardized processing, and then the calculation and optimization training data are performed, and in some other embodiments, other quality evaluation indexes can be selected to construct the quality evaluation matrix.
Referring to fig. 4, calculating, based on the standardized evaluation matrix B, the comprehensive cleaning quality index corresponding to each parameter training data set specifically includes:
Based on the quality evaluation matrix B, an idealized optimal quality matrix M + is screened out;
Based on a distance calculation formula, respectively calculating a vector distance between the silicon nitride layer residue standardized value and the wafer cleaning time-consuming standardized value of each parameter training data set and an idealized optimal quality matrix M +, and taking the vector distance as a comprehensive cleaning quality index corresponding to the parameter training data set.
The distance calculation formula is specifically as follows:
Wherein Z l is the comprehensive cleaning quality index of the 1 st parameter training data set.
It can be understood that the smaller the standardized value of the residual amount of the silicon nitride layer is, the higher the cleaning degree is, the smaller the standardized value of the time consuming for cleaning the wafer is, the smaller the time required for explaining the cleaning process is, however, in the normal processing state, the standardized value of the residual amount of the silicon nitride layer and the standardized value of the time consuming for cleaning the wafer cannot be simultaneously satisfied, the minimum value of the standardized value of the residual amount of the silicon nitride layer and the standardized value of the time consuming for cleaning the wafer are all reached, an ideal optimal quality matrix is formed, vector distances between the standardized value of the residual amount of the silicon nitride layer, the standardized value of the time consuming for cleaning the wafer and the ideal optimal quality matrix are calculated respectively, so as to obtain comprehensive cleaning quality indexes, and the smaller the comprehensive cleaning quality indexes are the corresponding cleaning quality and the efficiency are the more solved the ideal optimal state.
Referring to fig. 5, the calculation to obtain the cleaning parameters of the silicon nitride layer according to the optimized training data specifically includes:
Extracting historical parameter values corresponding to each silicon nitride layer cleaning parameter in the optimized training data to serve as silicon nitride layer cleaning parameter training data;
Calculating cleaning parameters of the silicon nitride layer through a parameter calculation formula based on the training data of the cleaning parameters of the silicon nitride layer;
The parameter calculation formula is:
Wherein, W g is the parameter value of the cleaning parameter of the g-th silicon nitride layer, and W gu is the history parameter value corresponding to the cleaning parameter of the g-th silicon nitride layer in the u-th optimized training data.
In the scheme, the silicon oxide layer is cleaned by taking the average value of a plurality of cleaning process parameters close to the optimal ideal state as the final cleaning parameter, so that the cleaning defect caused by calculation errors can be effectively reduced while the comprehensive cleaning quality of the silicon oxide layer is ensured, the stability in the wafer manufacturing process is greatly improved, and the defective products in wafer processing are reduced.
In summary, the invention has the advantages that: the intelligent analysis is carried out by combining the processing data of the wafer cleaning process, and the optimized processing parameter value of the wafer cleaning process can be automatically set after the silicon nitride layer is processed, so that the stability and the reliability of the whole process flow are ensured, the cleaning quality of the wafer can be effectively ensured, and the subsequent processing yield of the wafer is obviously improved.
The foregoing has shown and described the basic principles, principal features and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, and that the above embodiments and descriptions are merely illustrative of the principles of the present invention, and various changes and modifications may be made therein without departing from the spirit and scope of the invention, which is defined by the appended claims. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (8)
1. A method for cleaning a silicon part, comprising:
Constructing a parameter training database, wherein the parameter training database is used for storing historical processing parameters and historical cleaning data of a silicon nitride layer of a wafer;
acquiring processing parameters of a silicon nitride layer of a wafer, and recording the processing parameters as real-time processing parameters of the silicon nitride layer of the wafer, wherein the processing parameters of the silicon nitride layer of the wafer at least comprise deposition process parameters and annealing process parameters;
According to the real-time processing parameters of the silicon nitride layer of the wafer, a plurality of parameter training data sets are called from a parameter training database;
screening a plurality of optimized training data from the parameter training data set;
calculating according to the optimized training data to obtain cleaning parameters of the silicon nitride layer;
performing a silicon nitride layer cleaning process according to the silicon nitride layer cleaning parameters;
acquiring cleaning processing data under the current silicon nitride layer cleaning process, and adding the cleaning processing data under the current silicon nitride layer cleaning process into a parameter training database to obtain an updated parameter training database;
and when new cleaning of the silicon nitride layer is carried out, calculating cleaning parameters of the silicon nitride layer by adopting the updated parameter training database.
2. The method for cleaning a silicon part according to claim 1, wherein the step of retrieving a plurality of parameter training data sets corresponding to the current silicon nitride layer from the parameter training database according to the current processing parameters of the silicon nitride layer comprises:
Calculating the similarity between the real-time processing parameters of the silicon nitride layers of the wafers and the historical processing parameters of the silicon nitride layers of each wafer respectively through a processing similarity calculation formula;
Screening a plurality of historical processing parameters of the wafer silicon nitride layer with the similarity smaller than a preset value, and marking the historical processing parameters as fitting processing parameters of the wafer silicon nitride layer;
And taking historical cleaning data corresponding to the fitting processing parameters of the silicon nitride layer of the wafer as a parameter training data set.
3. The method for cleaning a silicon part according to claim 2, wherein the process similarity calculation formula is specifically:
Wherein L a is the similarity between the real-time processing parameters of the silicon nitride layer of the wafer and the historical processing parameters of the silicon nitride layer of the wafer a, x i0 is the i deposition process parameter value in the real-time processing parameters of the silicon nitride layer of the wafer a, x ia is the i deposition process parameter value in the historical processing parameters of the silicon nitride layer of the wafer a, n is the total number of deposition process parameters, y j0 is the j annealing process parameter value in the real-time processing parameters of the silicon nitride layer of the wafer a, y ja is the j annealing process parameter value in the historical processing parameters of the silicon nitride layer of the wafer a, and m is the total number of annealing process parameters.
4. A method for cleaning a silicon part according to claim 3, wherein the screening the optimized training data from the parameter training data set comprises:
determining a plurality of quality evaluation indexes of wafer cleaning, wherein the quality evaluation indexes at least comprise residual quantity of a silicon nitride layer and time consumption of wafer cleaning;
Extracting the residual quantity of the silicon nitride layer and the time consumption for cleaning the wafer from the parameter training data set to form a quality evaluation matrix A;
Carrying out standardization processing on the quality evaluation matrix A to obtain a standardization evaluation matrix B;
Based on the standardized evaluation matrix B, calculating a comprehensive cleaning quality index corresponding to each parameter training data set;
Setting an optimized training total number N, screening out parameter training data sets with the top N bits of comprehensive cleaning quality indexes according to the sequence from small to large, and taking the parameter training data sets as optimized training data.
5. The method for cleaning a silicon part according to claim 4, wherein the normalizing the quality evaluation matrix a to obtain a normalized evaluation matrix B specifically comprises:
the quality evaluation matrix Wherein a l1 is the residual quantity of the silicon nitride layer of the 1 st parameter training data set clock, a l2 is the time consumption of wafer cleaning of the 1 st parameter training data set clock, and k is the total number of parameter training data sets;
Respectively calculating a silicon nitride layer residue standardized value and a wafer cleaning time-consuming standardized value through a standardized formula;
The standardized formula is:
Wherein b l1 is a standardized value of the residual quantity of the silicon nitride layer in the 1 st parameter training data set, and b l2 is a standardized value of the time consumption of wafer cleaning in the 1 st parameter training data set;
A quality evaluation matrix B is constructed and is used for the construction of a quality evaluation matrix,
6. The method for cleaning a silicon part according to claim 5, wherein calculating the comprehensive cleaning quality index corresponding to each parameter training data set based on the standardized evaluation matrix B specifically comprises:
Based on the quality evaluation matrix B, an idealized optimal quality matrix M + is screened out;
Based on a distance calculation formula, respectively calculating a vector distance between the silicon nitride layer residue standardized value and the wafer cleaning time-consuming standardized value of each parameter training data set and an idealized optimal quality matrix M +, and taking the vector distance as a comprehensive cleaning quality index corresponding to the parameter training data set.
7. The method for cleaning a silicon part according to claim 6, wherein the distance calculation formula is specifically:
Wherein Z 1 is the comprehensive cleaning quality index of the 1 st parameter training data set.
8. The method for cleaning a silicon part according to claim 7, wherein the calculating to obtain the cleaning parameters of the silicon nitride layer according to the optimized training data specifically comprises:
Extracting historical parameter values corresponding to each silicon nitride layer cleaning parameter in the optimized training data to serve as silicon nitride layer cleaning parameter training data;
Calculating cleaning parameters of the silicon nitride layer through a parameter calculation formula based on the training data of the cleaning parameters of the silicon nitride layer;
the parameter calculation formula is as follows:
Wherein, W g is the parameter value of the cleaning parameter of the g-th silicon nitride layer, and W gu is the history parameter value corresponding to the cleaning parameter of the g-th silicon nitride layer in the u-th optimized training data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410351527.2A CN118230855A (en) | 2024-03-25 | 2024-03-25 | Wafer cleaning method for silicon part |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410351527.2A CN118230855A (en) | 2024-03-25 | 2024-03-25 | Wafer cleaning method for silicon part |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118230855A true CN118230855A (en) | 2024-06-21 |
Family
ID=91497516
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410351527.2A Pending CN118230855A (en) | 2024-03-25 | 2024-03-25 | Wafer cleaning method for silicon part |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118230855A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113130356A (en) * | 2021-04-15 | 2021-07-16 | 长江存储科技有限责任公司 | Control method and control device for wafer cleaning and computer readable storage medium |
| WO2023189170A1 (en) * | 2022-03-30 | 2023-10-05 | 株式会社荏原製作所 | Information processing device, inference device, machine-learning device, information processing method, inference method, and machine-learning method |
| CN117253775A (en) * | 2023-08-25 | 2023-12-19 | 江苏亚电科技有限公司 | Wafer cleaning method and device based on effect prediction |
-
2024
- 2024-03-25 CN CN202410351527.2A patent/CN118230855A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113130356A (en) * | 2021-04-15 | 2021-07-16 | 长江存储科技有限责任公司 | Control method and control device for wafer cleaning and computer readable storage medium |
| WO2023189170A1 (en) * | 2022-03-30 | 2023-10-05 | 株式会社荏原製作所 | Information processing device, inference device, machine-learning device, information processing method, inference method, and machine-learning method |
| CN117253775A (en) * | 2023-08-25 | 2023-12-19 | 江苏亚电科技有限公司 | Wafer cleaning method and device based on effect prediction |
Non-Patent Citations (1)
| Title |
|---|
| 蒋舒宇等: "基于Kohonen神经网络的晶圆光刻流程动态调度方法", 上海交通大学学报, vol. 43, no. 11, 28 November 2009 (2009-11-28), pages 1788 - 1793 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chien et al. | Manufacturing intelligence to forecast and reduce semiconductor cycle time | |
| CN106815462B (en) | Recognize the System and method for of the basic reason of yield loss | |
| CN110503288A (en) | Consider the System and method for of the interactive identification yield loss reason of board | |
| CN111105160A (en) | Steel quality prediction method based on tendency heterogeneous bagging algorithm | |
| CN107862415B (en) | Yield prediction method | |
| CN118414589A (en) | Manufacturing equipment part quality management system | |
| CN118230855A (en) | Wafer cleaning method for silicon part | |
| Fang et al. | Adagrad gradient descent method for ai image management | |
| CN111062539A (en) | Total electric quantity prediction method based on secondary electric quantity characteristic clustering analysis | |
| CN116842439A (en) | Semiconductor quality prediction method based on model | |
| US20230419107A1 (en) | Virtual metrology method based on convolutional autoencoder and transfer learning and system thereof | |
| CN112085298B (en) | Non-sequential flood frequency analysis method considering historical flood | |
| TWI234217B (en) | Method for analyzing wafer test parameters | |
| CN104537228B (en) | Glass furnace Study of Temperature Forecasting method based on residual error entropy and the double SVMs of layering | |
| CN116188877B (en) | Method and system for detecting and classifying unknown wafer defect categories | |
| CN108268987A (en) | The quality of Suresh Kumar product estimates method | |
| CN114969017B (en) | Wind power data cleaning method, cleaning device and prediction method | |
| TWI389233B (en) | In-line wafer measurement data compensate method | |
| CN111308977A (en) | Industrial process modeling method for sag and short-time interruption severity assessment and treatment | |
| CN116796806A (en) | Integrated circuit deposition film thickness characteristic processing method based on WOA-residual convolution neural network model | |
| JP2010140276A (en) | Polishing prediction evaluation device, polishing prediction evaluating method, polishing prediction evaluation program | |
| CN117549205B (en) | Quartz wafer polishing method | |
| TW202006568A (en) | Feedback system for improving process quality of product and method thereof | |
| CN117392440B (en) | Textile fabric retrieval method and system based on tissue structure and color classification | |
| CN115169617B (en) | Mold maintenance prediction model training method, mold maintenance prediction method and system |
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 |