Method for monitoring temperature of semiconductor process cavity

Abstract

The invention provides a method for monitoring the temperature of a semiconductor process cavity. The method comprises the following steps: S1, providing a monitoring process program for monitoring the condition of the process cavity; s2, selecting monitoring process temperature control output curves, calculating an output temperature average value avgs and a standard deviation sigma s after each step of each monitoring process temperature control output curve is stable, calculating a total average value avg0 and a standard deviation sigma 0 of each step of all the monitoring process temperature control output curves in the period based on each avgs, and calculating unstable time Ts to obtain sample data; s3, removing abnormal data in the sample data to obtain summarized data; s4, according to the summarized data and a calculation rule, the unstable time T, the maximum average value, the minimum average value, the average standard deviation and the maximum standard deviation of all the steps are obtained through calculation, and management and control data are obtained; and S5, setting a management and control rule according to the management and control data, and monitoring the temperature of the process cavity. High-sensitivity and low-false-alarm-rate monitoring of temperature control output in the production process is achieved.

Description

Technical Field

[0001] The present invention belongs to the field of semiconductor technology, and particularly relates to a method for monitoring the temperature of a semiconductor process chamber. Background Art

[0002] During the semiconductor production process, the temperature of each component in the process chamber will affect the process therein, so temperature control is very important. With the development of production technology, more and more components require precise temperature control, and the requirements for temperature control are also getting higher and higher. The active temperature control method is to feedback the data obtained by the temperature sensor to the temperature control system, and adjust the output of cooling and heating in real time to ensure that the component reaches the target temperature. Due to its position relationship, the measurement data of the temperature sensor needs to be regularly calibrated with the actual component and product temperatures. The temperature data monitored during the production process includes the data directly measuring the temperature control target, the data measured by the temperature sensor, and the output data of the temperature control system. Among them, the data directly measuring the temperature control target is directly accurate, but the non-real-time data has low timeliness; the temperature sensor data is intuitive, and most abnormalities can be monitored during actual production. However, when specific problems occur in itself or temperature conduction, system self-excitation error control output may occur, resulting in the sensor data meeting the standard but the actual temperature deviating; the output data of the temperature control system is indirect data, affected by multiple factors in the entire temperature control loop. In theory, a wider range of abnormalities can be monitored. In actual production, temperature abnormalities not found by the temperature sensor measurement data can be found through abnormalities in the output data of the temperature control system. However, due to its many influencing factors and large noise, it is difficult to directly formulate alarm rules.

[0003] In different steps of actual production, the temperature curve and the temperature control output curve will show almost simultaneous or fixed-delay changes, and there will be typical directivity. However, the temperature curve and the temperature control output do not have a direct proportional relationship. Different process rules set to the same temperature have different temperature control outputs due to different other conditions. In the first few seconds, both will show certain fluctuations and then tend to be stable. Even for the same process rule and the same chamber, different temperature control output curves will appear when producing different products. The noise caused by the differences often drowns out the differences in abnormal signals; the differences in the output curves between chambers and even within the same chamber during two different maintenance cycles are also large, and the above data will be mixed together, resulting in a very low signal-to-noise ratio and making it difficult to directly find abnormal signals. Currently, usually, alarm methods are formulated specifically after abnormalities occur, which requires analyzing data and specifying different process rules and product rules, lacking predictability and generality.

[0004] In summary, the output data of the temperature control system is indirect data with many influencing factors, and it is difficult to directly use it to capture production abnormalities. To solve the above problems, a method for monitoring the actual temperature of a semiconductor process chamber needs to be provided. Summary of the Invention

[0005] The present invention aims to solve all or part of the above-mentioned problems in the prior art, and provides a method for monitoring the temperature of a semiconductor process chamber. Without additional hardware and software, a temperature monitoring method with universality, predictability, low manpower requirements, high sensitivity, and low false alarm rate is established; actual production data is analyzed, control rules for automated management are set, the output temperature is monitored, and then the actual temperature condition of the process chamber is monitored, filling the loopholes in the temperature monitoring system caused by abnormal temperature sensors, etc., while meeting high sensitivity and low false alarm rate.

[0006] The present invention provides a method for monitoring the temperature of a semiconductor process chamber, including the following steps: S1: Provide a monitoring process program for monitoring the condition of the process chamber, and run the monitoring process program during the production process; S2: Select the monitoring process temperature control output curves, and calculate the average output temperature avg s and standard deviation σ s for each step after stabilization of each of the monitoring process temperature control output curves. Based on each avg s calculate the total average avg0 and standard deviation σ0 of all steps of all monitoring process temperature control output curves in this period, and calculate the unstable time T s , to obtain sample data; S3: Eliminate the abnormal data in the sample data to obtain summary data; S4: According to the summary data and calculation rules, calculate the unstable time T, maximum average avg max , minimum average avg min , average standard deviation σ avg , maximum standard deviation σ max for each step respectively, to obtain control data; S5: According to the control data, set control rules to monitor the temperature of the process chamber. Without adding hardware and software, a set of temperature monitoring methods is established: analyze actual production data, set control rules for automated management, monitor the output temperature, and then monitor the actual temperature condition of the process chamber, filling the loopholes in the temperature monitoring system caused by abnormal temperature sensors, etc., while having high sensitivity and low false alarm rate.

[0007] In step S1, the monitoring process program needs to meet the following requirements: it is used during normal production; it has steps within the temperature range of normal temperature control used in normal production; the source of the control wafer is fixed or it is a process program without wafers, and the determined process program that meets the requirements is used to detect the temperature of the process chamber. Ensure that the temperature monitoring function can operate effectively in the actual production environment and provide accurate and reliable data support.

[0008] In step S2, the selected temperature control output curve of the process chamber needs to meet the following conditions: a. The difference between the temperature control output curve of the process chamber and the average value of the output curves of the corresponding steps of the product temperature control output curve is less than 10%; b. The time after the temperature control output curve of the process chamber stabilizes is greater than 10 s, ensuring that the selected temperature control output curve of the process chamber meets the actual production requirements and can be used to generate control rules.

[0009] In step S2, the monitored process temperature control output curve can be selected from the monitored process temperature control output curve of a single chamber within a single maintenance cycle. The selected temperature control output curve is calculated for subsequent setting of control rules.

[0010] In step S2, the monitored process temperature control output curve can be selected from the monitored process temperature control output curves of all existing chambers (one or more) during one or more normal production and one or more maintenance cycles. By expanding the source of the temperature control output curve, the subsequent setting of control rules is usually more robust and reliable.

[0011] In step S3, the abnormal data screening method is as follows: Calculate the average value avg0 and the standard deviation σ0 of each step in the sample data, determine the upper and lower limits as avg0 ± 3σ0, and the data exceeding the upper and lower limits is the abnormal data. The abnormal data is removed to ensure the accuracy and reasonableness of the subsequent rule setting.

[0012] In step S4, the calculation rules include: Step a. Calculate the average value avg of each step in the summary data s and the standard deviation σ s1 , determine the multiple of σ s1 , and according to avg s1 ± 4σ s1 to ± 6σ s1 , calculate the upper and lower limits a; Step b. Determine the maximum and minimum values of avg of each step in the summary data s1 , and calculate that the maximum value increased by 5% and the minimum value decreased by 5% are the calculated upper and lower limits b; Step c. Compare the upper and lower limits a and the upper and lower limits b obtained in steps a and b with the avg s respectively, and select the group with the value closer to the avg s1 from the upper and lower limits a and the upper and lower limits b, and determine it as avg s1 and avg max ; and the avg min and the avg max need to cover the maximum and minimum values of the avg min . Select a method closer to the actual production situation to set avg s and avg max and avg min .

[0013] The calculation rules further include: the value of the unstable time T needs to cover the upper limit of the unstable time in the summary data and not exceed 5 s; the value of σ max needs to cover the upper limit of the σ0 value in the summary data. Determine reasonable values for determining control rules to accurately monitor the production temperature of the process chamber.

[0014] In step S5, the control rules include: the avg of a single curve needs to satisfy avg min < avg < avg max , otherwise stop production and give an alarm; the σ of a single curve needs to satisfy σ < σ max , otherwise stop production and give an alarm. Monitor the actual temperature condition of the production process chamber through the alarm rules of automated management, fill the loopholes in the temperature monitoring system caused by abnormal temperature sensors, etc., and at the same time have high sensitivity and low false alarm rate.

[0015] The control rules further include: calculate the average value avg of the first 3 temperature control output curves after each maintenance and take effect the new rules: determine the σ avg multiple, according to avgc ± 4 * σ avg to ± 6 * σ avg , determine the control range, and the average value avg of each subsequent single curve needs to be within the control range, otherwise stop production and give an alarm. For each maintenance, this rule is temporarily invalidated until new 3 temperature control output curves are generated. Generate new rules based on the actual production data in the same cycle to monitor the temperature of the process chamber in the same cycle, and it is easier to capture production anomalies.

[0016] Compared with the prior art, the beneficial effects of the present invention mainly include the following: provide a method for monitoring the temperature of a semiconductor process chamber, without adding additional hardware and software, use the output data of the temperature control system to establish a monitoring method with universality, predictability, low manpower requirements, high sensitivity, and low false alarm rate. According to the actual production data, establish control rules of automated management to monitor the output data of the temperature control system, and then monitor the actual temperature condition of the production process chamber, fill the loopholes in the temperature monitoring system caused by abnormal temperature sensors, etc., and at the same time meet high sensitivity and low false alarm rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments recorded in the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0018] Figure 1 A flowchart of a method for monitoring the temperature of a semiconductor process chamber provided by the present invention.

[0019] Figure 2 A schematic diagram of the process chamber temperature control output curve and the product temperature control output curve.

[0020] Figure 3 A schematic diagram of data analysis within a single maintenance cycle for a single chamber.

[0021] Figure 4 A schematic diagram of a method for eliminating abnormal data.

[0022] Figure 5 A schematic diagram of data analysis within a full maintenance cycle of normal production for all existing chambers.

[0023] Figure 6 A schematic diagram of data analysis within different maintenance cycles for a single chamber. Detailed implementation manners

[0024] The following description and drawings fully disclose specific implementation manners of the present invention, enabling those skilled in the art to practice them. Other implementation manners may include structural, logical, electrical, process, and other changes. The examples only represent possible variations. Unless explicitly required, separate components and functions are optional, and the order of operations may vary. Parts and features of some implementation manners may be included in or replace parts and features of other implementation manners.

[0025] Example 1

[0026] This example provides a method for monitoring the temperature of a semiconductor process chamber, and the specific steps are as Figure 1 shown.

[0027] S1: Provide a monitor recipe for monitoring the status of a process chamber and run the monitor recipe during the production process, subject to the following conditions: a. It is used during normal production, preferably a recipe with a high usage frequency; b. It has steps within the temperature range of normal temperature control used in normal production, at least approaching the upper and lower limits of the working temperature range respectively, with a difference of less than 10%; c. The source of the control wafers used is fixed, or it is a waferless process recipe. Recipes that are regularly used during normal production, regardless of whether the current purpose is production or monitoring, and meeting the above conditions, can have the temperature monitoring function added to be used as a monitor recipe. For example, the waferless auto clean (WAC) process recipe existing in the etching production chamber, which needs to be carried out once between each wafer and has different working temperatures, and there is no wafer in the etching chamber during the use process. If there is no existing available process recipe, a new monitor process flow needs to be established, including a new monitor recipe.

[0028] S2: Select the monitor process temperature control output curve. In this embodiment, the monitor process temperature control output curve is selected from the monitor process temperature control output curve of a single chamber within a single maintenance cycle. Calculate the average output temperature avg after stabilization for each step of each monitor process temperature control output curve respectively s , standard deviation σ s , based on each avg s Calculate the total average avg0 and standard deviation σ0 for each step of all monitor process temperature control output curves within this cycle and calculate the unstable time T s , to obtain sample data; as Figure 2 shown, the selected process chamber temperature control output curve needs to meet the following conditions: The difference between the average value of the output curves of the corresponding steps of the process chamber temperature control output curve and the product temperature control output curve is less than 10%; The time after the process chamber temperature control output curve stabilizes is greater than 10s. In this embodiment, taking the PM cycle x of chamber A within a single maintenance cycle of a single chamber as an example, calculate and summarize the avg s , σ s and T s for the low-temperature step 1 and high-temperature step 2 respectively, that is Figure 3 the part circled within the box.

[0029] S3: Eliminate the abnormal data in the sample data to obtain summary data; as Figure 4 shown, the abnormal data screening method is: Based on the average value avg0 and standard deviation σ0 of each step in the sample data, determine the upper and lower limits as avg0 ± 3σ0, and the data exceeding the upper and lower limits are abnormal data. The remaining sample data after eliminating the abnormal data is the summary data.

[0030] S4: Calculate the unstable time T, maximum average value avg, minimum average value avg, average standard deviation σ, and maximum standard deviation σ for each step respectively according to the summarized data and the calculation rules. The calculation rules include: 1) In step a, calculate the average value avg and standard deviation σ of avg for each step in the summarized data, determine the multiple of σ, and calculate the upper and lower limits a according to avg ± 4σ to ± 6σ; step b, determine the maximum and minimum values of avg for each step in the summarized data, and calculate the upper and lower limits b by increasing the maximum value by 5% and decreasing the minimum value by 5%; step c, compare the upper and lower limits a and b obtained in steps a and b with avg respectively, select the group with a value closer to avg among the upper and lower limits a and b, and determine it as avg and avg; and avg and avg need to cover the maximum and minimum values of avg. If the upper and lower limits a calculated in step a do not cover the maximum and minimum values of avg, select the values of the upper and lower limits b calculated in step b as avg and avg; it also includes: 2) The value of the unstable time T needs to cover the upper limit of the unstable time in the summarized data and not exceed 5 s. If the upper limit exceeds 5 s, set it according to 5 s; 3) σ is the average value of the standard deviations in the summarized data; 4) The value of σ needs to cover the upper limit of the standard deviation σ value in the summarized data. max and minimum average value avg min 、average standard deviation σ avg 、maximum standard deviation σ max 。 The calculation rules include: 1) In step a, calculate the average value avg of avg for each step in the summarized data respectively s and standard deviation σ s1 , determine the multiple of σ s1 , according to avg s1 ± 4σ s1 to ± 6σ s1 , calculate the upper and lower limits a; step b, determine the maximum and minimum values of avg for each step in the summarized data s1 , calculate the upper and lower limits b by increasing the maximum value by 5% and decreasing the minimum value by 5%; step c, compare the upper and lower limits a and b obtained in steps a and b with the avg s respectively, select the group with a value closer to the avg s1 among the upper and lower limits a and b, and determine it as avg s1 and avg max ; and avg min and avg max need to cover the maximum and minimum values of avg min , if the upper and lower limits a calculated in step a do not cover the maximum and minimum values of avg s , select the values of the upper and lower limits b calculated in step b as avg s and avg max ; also includes: 2) The value of the unstable time T needs to cover the upper limit of the unstable time in the summarized data and not exceed 5 s. If the upper limit exceeds 5 s, set it according to 5 s; 3) σ min is the average value of the standard deviations in the summarized data; 4) The value of σ avg needs to cover the upper limit of the standard deviation σ value in the summarized data max 。 s value.

[0031] S5: Set control rules according to the control data in step S4 to monitor the temperature of the process chamber. The control rules include: 1) The avg of a single curve needs to satisfy avg min < avg < avg max , otherwise stop production and alarm; 2) The σ of a single curve needs to satisfy σ < σ max , otherwise stop production and alarm; 3) Calculate the average value avg of the first 3 temperature control output curves after each maintenance cAnd add new rules. The new rule 4) is: Determine σ avg multiple. According to avgc ± 4 * σ avg to ± 6 * σ avg , determine the control range. The average value avg of each subsequent single curve must be within the control range, otherwise stop production and alarm. σ avg multiple is determined. Considering the importance degree of the production process of this production cavity and the sensitivity to the temperature of the process cavity, the relatively important and sensitive multiples are set lower. Each time of maintenance, this rule is temporarily invalid until new 3-pen temperature control output curves are generated. The temperature control output curves used to calculate avg c must meet control rule 1) and control rule 2). Update avg once after the 8th or 10th temperature control output curve is generated c .

[0032] Embodiment 2

[0033] This embodiment is another embodiment of a method for monitoring the temperature of a semiconductor process cavity. Different from Embodiment 1, in this embodiment, in step S2, the monitored process temperature control output curve of a single cavity in a single cycle is extended to obtain the process cavity temperature control output curves of all existing cavities in a full maintenance cycle of normal production. For example, on the basis of the PM cycle x of cavity A in Embodiment 1, extend and calculate the avg of the low-temperature step 1 and high-temperature step 2 of the PM cycle y of cavity B, the PM cycle z of cavity C, and the PM cycle α of cavity D s , σ s and T s , as Figure 5 the part circled in the box. The remaining steps are the same as those in Embodiment 1

[0034] Select the monitored process temperature control output curve. In this embodiment, the monitored process temperature control output curve is selected from the process cavity temperature control output curves of all existing cavities in a full maintenance cycle of normal production, and calculate the average output temperature avg after stabilization of each step of each monitored process temperature control output curve s , standard deviation σ s . Based on each avg s calculate the total average value avg0 and standard deviation σ0 of each step of all monitored process temperature control output curves in this cycle and calculate the unstable time T s , to obtain sample data; the selected process cavity temperature control output curve needs to meet the following conditions: the difference between the average value of the output curves of the corresponding steps of the process cavity temperature control output curve and the product temperature control output curve is less than 10%; the time after the process cavity temperature control output curve is stabilized is greater than 10 s

[0035] Embodiment 3

[0036] This embodiment is another embodiment of a method for monitoring the temperature of a semiconductor process chamber. Different from Embodiments 1 and 2, in this embodiment, in step S2, it can be extended to obtain the process chamber temperature control output curves of different maintenance cycles of a single chamber. For example, on the basis of the PM cycle x of chamber A in Embodiment 1, extend the calculation of the low-temperature step 1, the high-temperature step 2, avg s , σ s , and T s in the PM cycle x-1, the PM cycle x-2, and the PM cycle x-n of chamber A, as shown in the part circled in the box Figure 6 . The remaining steps are the same as those in Embodiment 1.

[0037] Select the monitored process temperature control output curve. In this embodiment, the monitored process temperature control output curve is selected from the process chamber temperature control output curves of different maintenance cycles of a single chamber, and calculate the average output temperature avg s , standard deviation σ s of each step after stabilization of each monitored process temperature control output curve. Based on each avg s , calculate the total average avg0 and standard deviation σ0 of each step of all monitored process temperature control output curves in this cycle and calculate the unstable time T s to obtain sample data; the selected process chamber temperature control output curve needs to meet the following conditions: the difference between the average value of the output curves of the corresponding steps of the process chamber temperature control output curve and the product temperature control output curve is less than 10%; the time after the process chamber temperature control output curve stabilizes is greater than 10 s.

[0038] It should be understood that some common English nouns or letters used in this application for the convenience of clear narration are only for exemplary reference rather than restrictive interpretation or specific usage, and the protection scope of this application should not be limited by their possible Chinese translations or specific letters. It should also 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 such actual relationship or order between these entities or operations.

Claims

1. A method for monitoring the temperature of a semiconductor process chamber, characterized in that, Including the following steps: S1: Provide a monitoring process program for monitoring the status of the process chamber, and run the monitoring process program during the production process; S2: Select the monitored process temperature control output curve, and calculate the average output temperature avg after stabilization of each step of each of the monitored process temperature control output curves s , standard deviation σ s , and based on each of the avg s calculate the total average avg0 and standard deviation σ0 of each step of all the monitored process temperature control output curves in this period and calculate the unstable time T s , to obtain sample data; S3: Eliminate the abnormal data in the sample data to obtain summary data; S4: Calculate the unstable time T, maximum average value avg, minimum average value avg, average standard deviation σ, and maximum standard deviation σ for each step respectively according to the summarized data and the calculation rules, to obtain the control data; max and minimum average value avg min and average standard deviation σ avg and maximum standard deviation σ max , to obtain the control data; S5: Set control rules according to the control data to monitor the temperature of the process chamber.

2. The method for monitoring the temperature of a semiconductor process chamber according to claim 1, wherein In step S1, the monitoring process program needs to meet the following conditions simultaneously: be used during normal production; have steps within the temperature range of normal temperature control used in normal production; the source of the control wafer is fixed or it is a process program without wafers.

3. The method for monitoring the temperature of a semiconductor process chamber according to claim 1, wherein In step S2, the selected monitoring process temperature control output curve needs to meet the following conditions: the average difference between the output curves of the corresponding steps of the monitoring process temperature control output curve and the product temperature control output curve is less than 10%; the time after the process chamber temperature control output curve stabilizes is greater than 10s.

4. The method for monitoring the temperature of a semiconductor process chamber according to claim 1, wherein, In step S2, the monitoring process temperature control output curve can be selected from the monitoring process temperature control output curve of a single chamber within a single maintenance cycle.

5. The method for monitoring the temperature of a semiconductor process chamber according to claim 1, wherein In step S2, the monitoring process temperature control output curve can be selected from the monitoring process temperature control output curves of one or more chambers within one or more maintenance cycles.

6. The method for monitoring the temperature of a semiconductor process chamber according to claim 1, wherein In step S3, the abnormal data elimination method is: based on the total average value avg0 and the standard deviation σ0 of each step in the sample data, determine the upper and lower limits as avg0 ± 3σ0, and the data exceeding the upper and lower limits is the abnormal data.

7. The method for monitoring the temperature of a semiconductor process chamber according to claim 1, wherein In step S4, the calculation rules include: Step a. Calculate the average value avg s of each step in the summary data s1 and the standard deviation σ s1 . Determine the multiple of σ s1 . According to avg s1 ±4σ s1 to ±6σ s1 , calculate the upper and lower limits a; Step b. Determine the maximum and minimum values of avg s of each step in the summary data. Calculate that the maximum value increased by 5% and the minimum value decreased by 5% are the calculated upper and lower limits b; Step c. Compare the upper and lower limits a and the upper and lower limits b obtained in steps a and b with the avg s1 respectively. Select the group with a value closer to the avg s1 from the upper and lower limits a and the upper and lower limits b, and determine it as avg max and avg min ; and the avg max and the avg min should cover the maximum and minimum values of the avg s .

8. The method for monitoring the temperature of a semiconductor process chamber according to claim 7, wherein, The calculation rule further includes: the value of the unstable time T needs to cover the upper limit of the unstable time in the summary data and not exceed 5 s; the σ max value needs to cover the upper limit of the σ s value in the summary data; σ avg is the average value of the standard deviation σ s in the summary data.

9. The method for monitoring the temperature of a semiconductor process chamber according to claim 7, wherein, In step S5, the control rules include: the avg of a single curve needs to satisfy avg min < avg < avg max , otherwise stop production and give an alarm; the σ of a single curve needs to satisfy σ < σ max , otherwise stop production and give an alarm.

10. The method for monitoring the temperature of a semiconductor process chamber according to claim 9, wherein, The control rules also include: calculating the average value avg of the first 3 temperature control output curves after each maintenance and taking effect the new rule: determining the σ avg multiple, according to avg c ±4*σ avg to ±6*σ avg , determining the control range, and the average value avg of each subsequent single curve must be within the control range, otherwise production is stopped and an alarm is given. For each maintenance, this rule is temporarily invalidated until new 3 temperature control output curves are generated.

Images