A method for constructing a wafer processing process control model

Abstract

The application discloses a method for constructing a wafer processing process control model, and belongs to the technical field of wafer processing production. The method collects cutting temperature field data, cutting data, grinding data, polishing data, cleaning data and surface profile data obtained after each process of a plurality of single wafers; according to the surface profile data obtained after the cutting process, expected cutting temperature field data and reference cutting data can be screened out, and the surface profile condition of a cutting wafer to be produced can be analyzed and predicted according to comparison between the expected cutting temperature field data and real-time monitoring temperature field data of the cutting wafer to be produced, so that a process control model capable of associating the surface profile of the cutting wafer with the cutting temperature field and the cutting process is obtained, instant feedback in the cutting process can be realized, the processing efficiency and the yield of the cutting process are improved, and qualified wafers can be produced in large quantities.

Description

Technical Field

[0001] This application relates to a method for constructing a wafer processing control model, belonging to the field of wafer processing and production technology. Background Technology

[0002] Semiconductors, with conductivity between that of conductors and insulators, are used to manufacture electronic materials for semiconductor devices and integrated circuits. Currently, the primary form of semiconductor used in electronic components is the wafer. Wafers are obtained from crystal rods through dicing, grinding, polishing, and cleaning processes. Each of these processes affects the wafer's surface area data to varying degrees.

[0003] Currently, the dicing process is the first step in wafer production, and therefore it is crucial to the surface quality of the wafer. However, the current dicing process still faces the following problems: First, due to the high hardness of semiconductor materials (generally greater than 6 on the Mohs scale), the wire dicing process has a long processing cycle. The entire dicing process cannot be monitored in real time, and if a quality problem occurs during processing, the entire batch produced will be substandard, leading to low processing efficiency and poor product yield control. Second, when problems are discovered during product inspection, it is impossible to effectively trace the source, let alone quickly prevent the same problem from occurring in subsequent processes. Third, the long processing cycle of the dicing process and the inability to provide real-time feedback on the processing status affect the controllability and effectiveness of the dicing process, thereby impacting the surface quality of the wafer.

[0004] The aforementioned problems result in the wafer dicing process still being blind dicing, leading to poor and uneven wafer quality after dicing. Subsequent grinding, polishing, and cleaning processes find it difficult to make large-scale adjustments to the surface quality based on this. Therefore, there is a lack of a control model that can provide real-time feedback and process adjustment for wafer processing. Summary of the Invention

[0005] To address the aforementioned issues, a method for constructing a wafer processing control model is provided. This model eliminates the influence of grinding, polishing, and cleaning processes on the wafer surface shape, combines wafer surface shape data, cutting temperature field data, and cutting data to establish a model relationship between surface shape, temperature field, and cutting. This model can provide real-time feedback on the processing status of the cutting process, effectively trace the source, improve the processing efficiency and yield of the cutting process, and facilitate the mass production of qualified wafers.

[0006] This application provides a method for constructing a wafer fabrication process control model, the method comprising the following steps:

[0007] S1. For a crystal rod of a predetermined specification, a cutting process, a grinding process, a polishing process, and a cleaning process are performed sequentially to obtain a wafer. Multiple single-wafer data pairs are collected to form a first data set. Each single-wafer data pair in the multiple single-wafer data pairs corresponds one-to-one with a single wafer and includes the cutting temperature field data, cutting data, grinding data, polishing data, cleaning data, and surface shape data obtained after each process for that single wafer.

[0008] S2. Keeping the grinding, polishing and cleaning processes unchanged, and using the surface data obtained after the single-wafer data alignment and cutting process as a condition that it falls within the preset expected cutting surface data range, a second data set is selected from the first data set. The cutting temperature field data of the second data set is used as the expected cutting temperature field data of a single wafer. The cutting data of the second data set is used to construct the reference cutting data of a single wafer.

[0009] S3. Form a process control model based on the comparison between the expected cutting temperature field data and the real-time monitoring temperature field data of the wafer to be cut during the cutting process, to analyze and predict the surface shape of the wafer to be cut, and to use the reference cutting data as a reference for adjusting the cutting parameters.

[0010] The existing method for adjusting the cutting process is to wait until the crystal rod is cut, and then inspect the cut wafers to obtain the corresponding surface shape before adjusting the cutting process parameters. Therefore, the process adjustment of the cut wafers is inefficient, time-consuming, and has a low yield. When the cut wafers of this batch are unqualified, there is no way to remedy the situation, resulting in the waste of the crystal rods. The only solution is to remedy the situation in the next batch of cut wafers.

[0011] The aforementioned construction method uses cutting temperature field data as a medium to establish a reliable link between cutting yield and cutting data. Then, by detecting the surface profile of the cut wafer, a link is established between the cut wafer and the cutting temperature field data, further establishing the relationship between cutting data, cutting temperature field data, and surface profile data. This allows for real-time feedback of the cutting data through the temperature field distribution in the cutting area, and this feedback can be verified through the final inspection results of the cut wafer. Furthermore, recording and storing the cutting temperature field data facilitates tracing the source of process anomalies that occur during the cutting process.

[0012] After establishing the model, the surface shape of the wafer to be cut can be analyzed and predicted based on the temperature field data of the cutting area. This can guide the real-time adjustment of the cutting data to obtain wafers that meet the surface shape requirements. This cutting process has high adjustment efficiency and short time consumption. Furthermore, because it is a real-time adjustment, it can ensure that the wafers in this batch can be corrected in time without waiting for subsequent feedback, thereby improving the yield of the cut wafers.

[0013] Optionally, the method further includes:

[0014] S4. Determine whether the cut wafer is qualified. For qualified cut wafers, keep the cutting process, polishing process and cleaning process unchanged. Use the surface data obtained after the single wafer data alignment grinding process falling into the preset expected grinding surface data range as a condition. Select a third data set from the second data set. Use the grinding data of the third data set to construct the reference grinding data of a single wafer. Use the reference grinding data as the reference for adjusting the grinding parameters in the process control model.

[0015] For defective cut wafers, determine whether the defect is due to slight deviation in face shape, minor edge chipping, or surface marks. If so, perform normal grinding on the cut wafers before reassessing their quality. Otherwise, directly classify them as defective. Slight deviation in face shape refers to a face shape value exceeding the acceptable face shape value by less than 5%, and minor edge chipping refers to an edge chipping size of less than 10μm.

[0016] Based on the judgment result in step S4, it is determined whether the diced wafer is qualified. Qualified diced wafers need to go through grinding, polishing and cleaning processes to obtain wafers. Therefore, the grinding data of the grinding process is still an important step in determining whether qualified wafers can be obtained. When the dicing, polishing and cleaning processes remain unchanged, reference grinding data can be determined to ensure the qualification of the ground wafer. If it is unqualified, it can be returned to the grinding process for reprocessing. Correction processing within the thickness range is carried out until it is qualified, and then subsequent processes can be carried out to prepare wafers, further improving the feedback efficiency of the model and reducing waste in the production process.

[0017] Optionally, the method further includes:

[0018] S5. Analyze the surface shape of the polished wafer produced in the polishing process, determine whether the polished wafer is qualified, and keep the cutting process, polishing process and cleaning process unchanged. Use the surface shape data obtained after the single wafer data is aligned with the polishing process as a condition that it falls within the preset expected polishing surface shape data range. Select a fourth data set from the third data set, construct reference polishing data for a single wafer with the polishing data of the fourth data set, and use the reference polishing data range as a process control model for adjusting polishing parameters.

[0019] For unqualified polished wafers, such as those with excessive surface shape, scratches, or particle contamination, they can be returned to the polishing process for reprocessing and correction within the thickness range. If the thickness or surface shape of the corrected polished wafer is still unqualified, it will be judged as an unqualified product. If the unqualification is caused by other factors such as edge chipping, cracks, or crystal defects, it will be directly judged as an unqualified product.

[0020] Optionally, the method further includes:

[0021] S6. Analyze the surface shape of the polished wafer produced in the polishing process, determine whether the polished wafer is qualified, and keep the cutting process, grinding process and polishing process unchanged for qualified polished wafers. Using the condition that the surface shape data obtained after the single-wafer data alignment and cleaning process falls within the preset expected cleaning surface shape data range, select the fifth data set from the fourth data set, construct the reference cleaning data for a single wafer with the cleaning data of the fifth data set, and use the reference cleaning data range as the reference for adjusting the cleaning parameters in the process control model.

[0022] Optionally, the dicing process has a weight of 60%-70% on the final wafer surface area data;

[0023] The grinding process has a weighting of 20%-30% on the final wafer surface area data;

[0024] The impact weight of the polishing process on the final wafer surface area data is 5%-15%;

[0025] The cleaning process has a weight of 0-1% on the final wafer surface area data.

[0026] The above model determines the influence weight of each processing step on the final wafer, so as to determine the optimal processing technology and thus construct a complete processing control model from crystal rod to wafer. Based on this model, the various processing steps of the wafer can be selectively controlled and adjusted to obtain wafers with the best surface quality.

[0027] Optionally, in step S3, the first real-time monitoring temperature field data of the wafer to be cut in the cutting process is compared with the expected cutting temperature field data to obtain a first comparison difference. The magnitude of the first comparison difference is used as the first basis for analyzing and predicting the surface shape of the wafer to be cut. For wafers to be cut that are determined to be unqualified by the first basis, the real-time cutting data of the wafer to be cut is adjusted.

[0028] The first real-time monitored cutting temperature field data can be used as a basis for judging whether the surface shape of the cut wafer is qualified. Therefore, the difference between the cutting temperature field data during the cutting process and the expected cutting temperature field data can be determined based on the first comparison difference, and the surface shape of the cut wafer can be judged based on the difference.

[0029] By using the changes in the first comparison difference to infer the cutting conditions, the processing data of unqualified wafers to be cut can be adjusted in a timely manner, thereby making the cutting process adjustment visible and improving cutting efficiency and yield.

[0030] Optionally, the face data includes at least one of Bow, Warp, TTV, and LTV;

[0031] The processing data includes at least one of the following: cutting fluid flow rate, cutting fluid temperature, cutting line speed, cutting line tension, auxiliary material concentration, workpiece feed speed, and swing angle.

[0032] The grinding data includes at least one of the following: grinding fluid flow rate, grinding fluid temperature, grinding fluid mass concentration, relative rotation speed, and pressure.

[0033] The polishing data includes at least one of the following: polishing fluid flow rate, polishing fluid temperature, polishing fluid mass concentration, relative rotation speed, and pressure.

[0034] The cleaning data includes at least one of the following: cleaning fluid flow rate, cleaning fluid temperature, and cleaning time.

[0035] Optionally, step S3 further includes: real-time monitoring of the temperature field of the wafer to be cut after adjusting the instantaneous processing data, obtaining second real-time monitored cutting temperature field data, comparing the second real-time monitored cutting temperature field data with the expected cutting temperature field data, and obtaining a second comparison difference, using the magnitude of the second comparison difference as a second basis for whether the surface shape of the wafer to be cut is qualified.

[0036] The cutting temperature field data after timely adjustment of the processing data is analyzed to obtain a second comparison difference. This second comparison difference, based on the first comparison difference, determines the difference between the adjusted cutting temperature field data and the expected cutting temperature field data, thus analyzing whether the produced wafer is qualified. This second comparison difference can further determine the surface shape of the produced wafer. If the second comparison difference is too large, exceeding the control range of the process control model, the produced wafer can be determined to be in an unrecoverable state, and the cutting process can be stopped in time to avoid wasting cutting time, allowing the next batch of crystal rods to be cut. If the magnitude of the second comparison difference is within the acceptable range for the surface shape of the produced wafer, it proves that the adjustment of the cutting data is effective, and qualified wafers can be obtained.

[0037] Furthermore, based on the differences between the first and second comparison differences, the influence weight of the cutting data on the cutting process can be determined, thereby guiding the model to determine the adjustment order of the cutting data according to the magnitude of the first comparison difference. For example, when the first comparison difference is large, that is, when the real-time monitored cutting temperature field data differs significantly from the expected cutting temperature field data, the cutting data with a larger influence weight can be directly adjusted to ensure that the second comparison difference falls within a suitable range, so that the real-time monitored cutting temperature field data falls within the range of the expected cutting temperature field data. When the first comparison difference is small, the cutting data with a smaller influence weight can be adjusted to avoid the second comparison difference changing too much compared to the first comparison difference, thereby adjusting the real-time monitored cutting temperature field data within a small range so that the real-time monitored cutting temperature field data falls within the range of the expected cutting temperature field data.

[0038] Optionally, the weighting levels of the influence of the cutting data on the final wafer surface area data are as follows: the first level includes cutting fluid temperature, workpiece feed speed, and cutting fluid flow rate; the second level includes cutting line speed and auxiliary material mass concentration; and the third level includes swing angle and cutting line tension.

[0039] The weighting levels of the impact of the polishing data on the final wafer surface area data are as follows: the first level includes polishing slurry temperature and polishing slurry mass concentration, and the second level includes relative rotation speed, pressure, and polishing slurry flow rate.

[0040] The weighting levels of the impact of the polishing data on the final wafer surface shape data are as follows: the first level includes relative rotation speed and polishing slurry temperature, and the second level includes polishing slurry mass concentration, pressure, and polishing slurry flow rate.

[0041] The weighting levels of the influence of the cleaning data on the final wafer surface area data are as follows: the first level includes the flow rate of the polishing slurry and the temperature of the cleaning slurry, and the second level includes the cleaning time.

[0042] Optionally, the type of cutting data can be adjusted: first adjust the workpiece feed speed and cutting fluid flow rate, then adjust the cutting line speed and swing angle, and then adjust the cutting line tension, cutting fluid temperature and auxiliary material concentration.

[0043] Adjustment methods for cutting data: When the surface shape deviates significantly from the reference range, adjust at least one of the following: workpiece feed speed, cutting fluid flow rate, and cutting fluid temperature; when the surface shape deviates slightly from the reference range, adjust at least one of the following: cutting line tension, cutting line speed, swing angle, and auxiliary material concentration.

[0044] Adjust the type of grinding data: prioritize adjusting relative rotation speed, pressure, and grinding fluid flow rate, then adjust grinding fluid temperature and grinding fluid mass concentration;

[0045] Adjustment methods for grinding data: When the surface profile deviates significantly from the reference range, adjust the grinding slurry temperature and / or the grinding slurry concentration; when the surface profile deviates slightly from the reference range, adjust at least one of the relative rotation speed, pressure, and grinding slurry flow rate.

[0046] Adjust the type of polishing data: prioritize adjusting pressure, relative rotation speed, and polishing fluid flow rate, then adjust polishing fluid mass concentration and polishing fluid temperature;

[0047] Polishing data adjustment methods: When the surface shape deviates significantly from the reference range, adjust the relative rotation speed and / or polishing slurry temperature; when the surface shape deviates slightly from the reference range, adjust at least one of the following: polishing slurry concentration, pressure, and polishing slurry flow rate.

[0048] Adjust the type of cleaning data: prioritize adjusting the cleaning fluid flow rate and cleaning time, then adjust the cleaning fluid temperature;

[0049] Adjustment method for cleaning data: When the surface pattern deviates significantly from the reference range, adjust the cleaning fluid flow rate and / or cleaning fluid temperature; when the surface pattern deviates slightly from the reference range, adjust the cleaning fluid flow rate and / or cleaning time.

[0050] Optionally, the cutting temperature field data includes temperature distribution information based on the cross-section of the crystal rod, wherein the temperature distribution information is grid-space distributed temperature data or image-based visualized temperature distribution information.

[0051] Due to the different cutting positions, even with the same processing data, the temperature field data at different locations on the crystal rod cross-section (i.e., the cutting surface) will vary. Therefore, using the temperature distribution of the crystal rod cross-section as temperature field data can build a more systematic process control model, thereby controlling the cutting process more accurately.

[0052] This visual temperature distribution information is achieved through a temperature measurement system, which includes an infrared detector, an image processing unit, and a monitor. The infrared detector obtains basic information about the temperature field at the cutting location, providing detailed temperature distribution data for the cutting area. The data detected by the infrared detector is transmitted to the image processor for processing, and finally, the visual temperature distribution information is presented on the monitor. This visual temperature distribution information can be displayed on-site and also remotely in real time, enabling online monitoring of the temperature field distribution at the cutting location.

[0053] Optionally, step S2 further includes constructing an upper threshold and a lower threshold corresponding to the expected cutting temperature field data. The temperature range formed by the upper threshold and the lower threshold is the temperature range of the expected cutting temperature field data. When the real-time monitored cutting temperature field data exceeds the upper threshold or falls below the lower threshold, the surface shape of the cut wafer is determined to be unqualified, thereby allowing the cutting data to be adjusted based on the reference cutting data range.

[0054] The beneficial effects of this application include, but are not limited to:

[0055] 1. The method for constructing a wafer processing control model according to this application eliminates the influence of grinding, polishing and cleaning processes on wafer surface shape, combines wafer surface shape data, cutting temperature field data and cutting data to establish a model relationship between surface shape, temperature field and cutting, which can provide real-time feedback on the processing status of the cutting process, effectively trace the source, and improve the processing efficiency and yield of the cutting process.

[0056] 2. According to the method of constructing a wafer processing control model of this application, after the model is established, the surface shape of the wafer to be produced can be analyzed and predicted based on the temperature field data of the cutting area. This can guide the cutting data to be adjusted in real time within the reference cutting data range, ensuring that the wafers of this cutting batch can be remedied in a timely manner without waiting for subsequent feedback, thereby improving the yield of the wafers.

[0057] 3. Based on the method for constructing a wafer processing control model in this application, the dicing, grinding, polishing and cleaning processes are screened and analyzed in sequence to determine the influence weight of each process on the wafer surface data, and the processing technology of the next process can be adjusted in a timely manner based on the surface data obtained from the previous process.

[0058] 4. According to the method for constructing a wafer processing control model of this application, the influence weight of each cutting data on the cutting process can be determined based on the difference between the first comparison difference and the second comparison difference, thereby guiding the model to determine the adjustment order of each parameter in the cutting data according to the magnitude of the first comparison difference, and improving the controllability of the model. Attached Figure Description

[0059] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0060] Figure 1 This is a schematic diagram illustrating the process of adjusting real-time cutting data based on the first real-time monitoring temperature field data of the cutting area, as described in the embodiments of this application. Detailed Implementation

[0061] The present application is described in detail below with reference to the embodiments, but the present application is not limited to these embodiments.

[0062] Unless otherwise specified, all raw materials used in the embodiments of this application were purchased through commercial channels.

[0063] The following embodiments use silicon carbide crystal rods as an example for illustration. Those skilled in the art will understand that the method for constructing a wafer processing control model can also be applied to crystal rods of other materials, such as silicon, germanium, gallium arsenide, ceramics, glass, sapphire, etc. The silicon carbide crystal rods used in the following embodiments do not constitute a limitation on this application.

[0064] Example

[0065] This embodiment relates to a method for constructing a wafer fabrication process control model, the method comprising the following steps:

[0066] S1. A silicon carbide crystal rod is sequentially subjected to a cutting process, a grinding process, a polishing process, and a cleaning process to obtain a wafer. Multiple single-wafer data pairs are collected to form a first data set. Each single-wafer data pair in the multiple single-wafer data pairs corresponds one-to-one with a single wafer and includes the surface shape data, cutting temperature field data, cutting data, grinding data, polishing data, and cleaning data of the single wafer.

[0067] S2. Keeping the grinding, polishing, and cleaning processes unchanged, determine the cutting data for the cutting process, including cutting fluid flow rate, cutting fluid temperature, cutting line speed, cutting line tension, auxiliary material concentration, workpiece feed speed, and swing angle. Monitor and collect the temperature field data of a single wafer in the cutting process in real time to obtain the grid spatial distribution temperature data or image-based visualized temperature distribution information of the crystal rod cross-section. Obtain the surface shape data by detecting the surface shape of a single wafer. Adjust the above cutting data, statistically analyze the temperature data of the cutting area under different cutting data, and obtain different surface shape data respectively.

[0068] In the first data set, the second data set is selected based on the condition that the surface data in the single data pair falls within the preset expected surface data range. The cutting temperature field data of the second data set is used as the expected cutting temperature field data of a single wafer. The cutting data of the second data set is used to construct the reference cutting data of a single wafer.

[0069] S3. Real-time detection of the temperature field in the crystal rod cutting area is performed to form a comparison between the expected cutting temperature field data and the real-time monitoring temperature field data of the wafer to be cut, so as to analyze and predict the surface shape of the wafer to be cut, and to adjust the process control model based on the reference cutting data.

[0070] According to step S2 above, the silicon carbide crystal rods to be cut are selected from the same batch of silicon carbide crystal rods with the same indicators. The specific cutting data is shown in Table 1. Cut wafers 1#-44# are obtained. The auxiliary materials are diamond powder, polyvinylpyrrolidone and polyethylene glycol, with a weight ratio of 100:1:2.

[0071] Table 1 Cutting Data

[0072]

[0073]

[0074]

[0075] The surface profile of wafers 1# to 44# was tested, and the test results are shown in Table 2. In the table, Bow represents curvature, Warp represents warpage, TTV represents total thickness deviation, and LTV represents local thickness deviation.

[0076] Table 2. Surface profile data of the diced wafers

[0077]

[0078]

[0079]

[0080] In Table 2, the Bow value of the cut wafer #10 is 8.3 μm, the Warp value is 18.5 μm, the TTV value is 8.2 μm, and the LTV value is 6.1 μm. These values ​​are used as the expected cut surface profile data for this embodiment. Therefore, the temperature range of the expected temperature field data is determined to be 15–21 °C. Based on the cut data in Table 1, the reference processing data for the cutting fluid flow rate is determined to be 2000 kg / h, the reference processing data for the cutting fluid temperature is 10 °C, the reference processing data for the cutting line speed is 25 m / s, the reference processing data for the cutting line tension is 25 N, the reference processing data for the auxiliary material mass concentration is 10%, the reference processing data for the workpiece feed speed is 20 μm / min, and the reference processing data range for the swing angle is 2°.

[0081] As a preferred embodiment, since the temperature field data is grid-spaced temperature data or image-based visualized temperature distribution information, the cross-section of the crystal rod can be divided into four regions from top to bottom. The width of each region accounts for 25% of the diameter of the crystal rod. The expected temperature range of the temperature field data in the first region is 15-19°C, the expected temperature range of the temperature field data in the second region is 17-21°C, the expected temperature range of the temperature field data in the third region is 17-21°C, and the expected temperature range of the temperature field data in the fourth region is 15-19°C.

[0082] Taking wafers 16# to 44# as examples, as qualified wafers, wafers 16# to 44# are subjected to a grinding process. The grinding slurry uses diamond powder with W50 of 4μm and water as the solvent, with 2% suspending agent added. The mass concentration of the grinding slurry in the table below refers to the mass concentration of the diamond powder. The specific grinding data is shown in Table 3, and wafers 1# to 29# are obtained. The specific differences are shown in Table 3 below.

[0083] Table 3 Grinding data

[0084]

[0085]

[0086]

[0087] The surface profile of the polished wafers 1#-29# was tested, and the test results are shown in Table 4.

[0088] Table 4. Surface profile data of the polished wafers

[0089]

[0090]

[0091] Table 4 shows that the Bow value of polished wafer #10 is 3.5 μm, Warp value is 8.4 μm, TTV value is 5.3 μm, and LTV value is 3.6 μm. These values ​​are used as the desired polishing surface profile data for this embodiment. Therefore, the reference polishing data for the polishing slurry flow rate is determined to be 10 L / min, the reference polishing slurry temperature is 25 °C, the reference polishing slurry mass concentration is 9%, the reference polishing speed is 15 rpm, and the reference polishing pressure is 100 g / cm³. 2 .

[0092] Taking polishing wafers 12 to 29# as an example, as qualified cut wafers, polishing wafers 12 to 29# is performed. The polishing fluid uses diamond powder with W50 of 1μm and water as the solvent, with 2% suspending agent added. The mass concentration of polishing fluid in the table below refers to the mass concentration of diamond powder. The specific polishing data is shown in Table 3, and polished wafers 1# to 18# are obtained. The specific differences are shown in Table 5 below.

[0093] Table 5 Polishing Data

[0094]

[0095]

[0096] The surface profile of polished wafers 1#-18# was tested, and the test results are shown in Table 6.

[0097] Table 6. Surface profile data of polished wafers

[0098]

[0099]

[0100] Table 6 shows that the Bow value of polished wafer #7 is 1.4 μm, Warp value is 3 μm, TTV value is 3 μm, and LTV value is 1.6 μm. These values ​​are used as the desired polishing surface profile data for this embodiment. Therefore, the reference polishing data for the polishing slurry flow rate is determined to be 5 L / min, the reference polishing slurry temperature is 25°C, the reference polishing slurry mass concentration is 10%, the reference polishing speed is 15 rpm, and the reference polishing pressure is 50 g / cm³. 2 .

[0101] Taking polished wafers 12# to 18# as examples, as qualified polished wafers, polished wafers 12# to 18# are cleaned by first cleaning with 5% ammonia water and then washing with water. The specific cleaning data is shown in Table 7, and wafers 1# to 7# are obtained. The specific differences are shown in Table 7 below.

[0102] Table 7 Cleaning Data

[0103] sample <![CDATA[Cleaning liquid flow rate (m 3 / h)]]> Cleaning fluid temperature (°C) Cleaning time (min) Wafer #1 1 25 60 Wafer #2 0.1 25 60 Wafer #3 5 25 60 Wafer #4 1 35 60 Wafer #5 1 15 60 Wafer #6 1 25 30 Wafer #7 1 25 120

[0104] The surface profile of wafers 1#-7# was tested, and the test results are shown in Table 8.

[0105] Table 8 Wafer Surface Data

[0106] sample Bow(μm) Warp(μm) TTV(μm) LTV(μm) Wafer #1 1.4 3.2 2.5 1.3 Wafer #2 1.5 3.2 2.4 1.1 Wafer #3 1.4 3.4 2.5 1.3 Wafer #4 1.1 3.3 2.4 1.3 Wafer #5 1.4 3.1 2.2 1.2 Wafer #6 1.5 3.3 2.3 1.3 Wafer #7 1.4 2.9 2.1 1.2

[0107] Table 8 shows that wafer #1 has a Bow value of 1.4 μm, a Warp value of 3.2 μm, a TTV value of 2.5 μm, and an LTV value of 1.3 μm. These values ​​are used as the desired cleaning surface profile data for this embodiment. Therefore, the reference cleaning data for the cleaning fluid flow rate is determined to be 1 m³ / min. 3 The reference cleaning data for cleaning fluid temperature is 25℃ / h and the reference cleaning time is 60min.

[0108] The LTV in the table above was measured using the light interferometry method and represents the overall average LTV. Based on Tables 1-8 above, reference cutting data, reference grinding data, reference polishing data, and reference cleaning data can be determined respectively, thereby constructing a wafer processing control model and providing a basis for the mass production of qualified wafers.

[0109] The influence weights of the dicing, grinding, polishing, and cleaning processes on the final wafer surface shape data were determined by comparing the impact of changes in different processing parameters on the magnitude of surface shape changes. Specifically, in the dicing process, the influence weight levels of the cutting fluid flow rate, cutting fluid temperature, cutting line speed, cutting line tension, auxiliary material concentration, workpiece feed rate, and tilt angle on the final wafer surface shape data were determined as follows: Level 1 includes cutting fluid temperature, workpiece feed rate, and cutting fluid flow rate; Level 2 includes cutting line speed and auxiliary material concentration; and Level 3 includes tilt angle and cutting line tension. Similarly, in the grinding process, the influence weight levels of grinding data on the final wafer surface shape data were determined as follows: Level 1 includes grinding fluid temperature and grinding fluid concentration; and Level 2 includes relative rotation speed, pressure, and grinding fluid flow rate. Finally, in the polishing process, the influence weight levels of polishing data on the final wafer surface shape data were determined as follows: Level 1 includes relative rotation speed and polishing fluid temperature; and Level 2 includes polishing fluid concentration, pressure, and polishing fluid flow rate. The weighting levels of the impact of cleaning data on the final wafer surface area data during the cleaning process are as follows: the first level includes the flow rate of polishing slurry and cleaning slurry, and the temperature of cleaning slurry; the second level includes the cleaning time.

[0110] As one implementation method, in step S3, refer to Figure 1 The first real-time monitoring temperature field data of the wafer to be cut during the cutting process is compared with the expected cutting temperature field data to obtain the first comparison difference. The magnitude of the first comparison difference is used as the first basis for analyzing and predicting the surface shape of the wafer to be cut. For wafers to be cut that are determined to be unqualified by the first basis, the real-time cutting data of the wafer to be cut is adjusted.

[0111] This model is dynamically adjustable. Based on the magnitude of the first comparison difference, it can analyze the surface data of the cut wafer caused by the cutting process in a timely manner, thereby adjusting the corresponding cutting data in a timely manner and realizing real-time feedback of the cutting status. This improves the model's process control capability and control effect, reduces the number of defective cut wafers, and improves the controllability of product yield.

[0112] As one implementation method, the specific method for adjusting the real-time cutting data of the wafer to be produced is as follows:

[0113] Adjust the type of cutting data: Prioritize adjusting the workpiece feed speed and cutting fluid flow rate, then adjust the cutting line speed and swing angle, and finally adjust the cutting line tension, cutting fluid temperature, and auxiliary material concentration.

[0114] Adjustment method for cutting data: When the first comparison difference is large, that is, when the surface shape result deviates greatly from the reference range, adjust at least one of the following: workpiece feed speed, cutting fluid flow rate, and cutting fluid temperature; when the first comparison difference is small, that is, when the surface shape result deviates little from the reference range, adjust at least one of the following: cutting line tension, cutting line speed, swing angle, and auxiliary material mass concentration.

[0115] As one implementation method, step S3 further includes: real-time monitoring of the temperature field of the wafer to be cut after adjusting the instant processing data, obtaining second real-time monitored cutting temperature field data, comparing the second real-time monitored cutting temperature field data with the expected cutting temperature field data, and obtaining a second comparison difference, using the magnitude of the second comparison difference as a second basis for whether the surface shape of the wafer to be cut is qualified.

[0116] For wafers to be cut that are deemed unqualified based on the first criterion, after the cutting method adjusts the real-time processing data in a timely manner, it is also necessary to obtain a second comparison difference through the second real-time monitoring data to determine whether the adjusted real-time processing data meets the production requirements of qualified wafers, thereby further improving the feedback effect of the cutting method and realizing the visibility of processing data adjustment.

[0117] The above are merely embodiments of this application, and the scope of protection of this application is not limited to these specific embodiments, but is determined by the claims of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the technical concept and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for constructing a wafer fabrication process control model, characterized in that, The method includes the following steps: S1. For a crystal rod of a predetermined specification, a cutting process, a grinding process, a polishing process, and a cleaning process are performed sequentially to obtain a wafer. Multiple single-wafer data pairs are collected to form a first data set. Each single-wafer data pair in the multiple single-wafer data pairs corresponds one-to-one with a single wafer and includes the cutting temperature field data, cutting data, grinding data, polishing data, cleaning data, and surface shape data obtained after each process for that single wafer. S2. Keeping the grinding, polishing and cleaning processes unchanged, and using the surface data obtained after the single-wafer data alignment and cutting process as a condition that it falls within the preset expected cutting surface data range, a second data set is selected from the first data set. The cutting temperature field data of the second data set is used as the expected cutting temperature field data of a single wafer. The cutting data of the second data set is used to construct the reference cutting data of a single wafer. S3. Form a process control model based on the comparison between the expected cutting temperature field data and the real-time monitoring temperature field data of the wafer to be cut during the cutting process, to analyze and predict the surface shape of the wafer to be cut, and to use the reference cutting data as a reference for adjusting the cutting parameters.

2. The method for constructing a wafer fabrication process control model according to claim 1, characterized in that, The method further includes: S4. Determine whether the cut wafer is qualified. For qualified cut wafers, keep the cutting process, polishing process and cleaning process unchanged. Use the surface data obtained after the single-wafer data alignment grinding process falling into the preset expected grinding surface data range as a condition. Select a third data set from the second data set. Use the grinding data of the third data set to construct the reference grinding data of a single wafer. Use the reference grinding data as the reference for adjusting the grinding parameters in the process control model.

3. The method for constructing a wafer fabrication process control model according to claim 2, characterized in that, The method further includes: S5. Analyze the surface shape of the polished wafer produced in the polishing process, determine whether the polished wafer is qualified, and keep the cutting process, polishing process and cleaning process unchanged. Use the surface shape data obtained after the single wafer data is aligned with the polishing process as a condition that it falls within the preset expected polishing surface shape data range. Select a fourth data set from the third data set, construct reference polishing data for a single wafer with the polishing data of the fourth data set, and use the reference polishing data as a reference for adjusting the polishing parameters in the process control model.

4. The method for constructing a wafer fabrication process control model according to claim 3, characterized in that, The method further includes: S6. Analyze the surface shape of the polished wafer produced in the polishing process, determine whether the polished wafer is qualified, and keep the cutting process, grinding process and polishing process unchanged for qualified polished wafers. Using the condition that the surface shape data obtained after the single-wafer data alignment and cleaning process falls within the preset expected cleaning surface shape data range, select the fifth data set from the fourth data set, construct the reference cleaning data for a single wafer with the cleaning data of the fifth data set, and use the reference cleaning data range as the reference for adjusting the cleaning parameters in the process control model.

5. The method for constructing a wafer fabrication process control model according to claim 1, characterized in that, The dicing process has a weighting of 60%-70% on the final wafer surface area data; The grinding process has a weighting of 20%-30% on the final wafer surface area data; The impact of the polishing process on the final wafer surface area data is 5%-15%; The cleaning process has a weight of 0-1% on the final wafer surface area data.

6. The method for constructing a wafer fabrication process control model according to claim 5, characterized in that, The face data includes at least one of Bow, Warp, TTV, and LTV; The cutting data includes at least one of the following: cutting fluid flow rate, cutting fluid temperature, cutting line speed, cutting line tension, auxiliary material concentration, workpiece feed speed, and swing angle. The grinding data includes at least one of the following: grinding fluid flow rate, grinding fluid temperature, grinding fluid mass concentration, relative rotation speed, and pressure. The polishing data includes at least one of the following: polishing fluid flow rate, polishing fluid temperature, polishing fluid mass concentration, relative rotation speed, and pressure. The cleaning data includes at least one of the following: cleaning fluid flow rate, cleaning fluid temperature, and cleaning time.

7. The method for constructing a wafer fabrication process control model according to claim 6, characterized in that, The weighting levels of the influence of the cutting data on the final wafer surface area data are as follows: the first level includes cutting fluid temperature, workpiece feed speed, and cutting fluid flow rate; the second level includes cutting line speed and auxiliary material concentration; and the third level includes swing angle and cutting line tension. The weighting levels of the impact of the polishing data on the final wafer surface area data are as follows: the first level includes polishing slurry temperature and polishing slurry mass concentration, and the second level includes relative rotation speed, pressure, and polishing slurry flow rate. The weighting levels of the impact of the polishing data on the final wafer surface shape data are as follows: the first level includes relative rotation speed and polishing slurry temperature, and the second level includes polishing slurry mass concentration, pressure, and polishing slurry flow rate. The weighting levels of the influence of the cleaning data on the final wafer surface area data are as follows: the first level includes the flow rate of the polishing slurry and the temperature of the cleaning slurry, and the second level includes the cleaning time.

8. The method for constructing a wafer fabrication process control model according to claim 7, characterized in that, Adjust the type of cutting data: Prioritize adjusting the workpiece feed speed and cutting fluid flow rate, then adjust the cutting line speed and swing angle, and finally adjust the cutting line tension, cutting fluid temperature, and auxiliary material concentration. Methods for adjusting cutting data: When the surface shape deviates significantly from the reference range, adjust at least one of the following: workpiece feed speed, cutting fluid flow rate, and cutting fluid temperature; when the surface shape deviates slightly from the reference range, adjust at least one of the following: cutting line tension, cutting line speed, swing angle, and auxiliary material concentration. Adjust the type of grinding data: prioritize adjusting relative rotation speed, pressure, and grinding fluid flow rate, then adjust grinding fluid temperature and grinding fluid mass concentration; Adjustment methods for grinding data: When the surface profile deviates significantly from the reference range, adjust the grinding slurry temperature and / or the grinding slurry concentration; when the surface profile deviates slightly from the reference range, adjust at least one of the relative rotation speed, pressure, and grinding slurry flow rate. Adjust the type of polishing data: prioritize adjusting pressure, relative rotation speed, and polishing fluid flow rate, then adjust polishing fluid mass concentration and polishing fluid temperature; Polishing data adjustment methods: When the surface shape deviates significantly from the reference range, adjust the relative rotation speed and / or polishing slurry temperature; when the surface shape deviates slightly from the reference range, adjust at least one of the following: polishing slurry concentration, pressure, and polishing slurry flow rate. Adjust the type of cleaning data: prioritize adjusting the cleaning fluid flow rate and cleaning time, then adjust the cleaning fluid temperature; Adjustment method for cleaning data: When the surface pattern deviates significantly from the reference range, adjust the cleaning fluid flow rate and / or cleaning fluid temperature; When the surface pattern deviates from the reference range by a small amount, adjust the cleaning fluid flow rate and / or cleaning time.

9. The method for constructing a wafer fabrication process control model according to claim 1, characterized in that, The cutting temperature field data includes temperature distribution information based on the cross-section of the crystal rod, which is either grid-space distributed temperature data or image-based visualized temperature distribution information.

Images