Information processing device, determination method, and water treatment system

Abstract

To discover early a defect of a semiconductor device.SOLUTION: An information processing device includes: an acquisition part 110 that acquires a liquid quality value of a waste liquid of a cleaning liquid used for cleaning an object; a reading part 120 that reads a reference liquid quality value serving as a reference from a database 400; a determination part 130 that determines whether a semiconductor device manufactured in a manufacturing step including a cleaning step of an object is a non-defective product based on the liquid quality value acquired by the acquisition part 110 and the reference liquid quality value read by the reading part 120; and a notification part 140 that notifies a determination result determined by the determination part 130.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to an information processing device, a determination method, and a water treatment system. [Background technology]

[0002] Whether a manufactured semiconductor device is a non-defective product or not is determined based on the results of a predetermined inspection of the semiconductor device. In the manufacturing of semiconductor devices, a method has been devised in which, in the process of removing a film stack on a semiconductor wafer by bringing a processing liquid into contact with the film, the processing liquid to be contacted is switched based on the concentration of substances in the processing liquid (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2006-120708 Summary of the Invention [Problem to be solved by the invention]

[0004] As mentioned above, it takes several months to conduct quality inspections of semiconductor devices and obtain a judgment result. Therefore, if a defective product is manufactured in the early stages of the manufacturing process, the defective product will remain on the production line for several months, resulting in a decrease in yield. Furthermore, the method described in Patent Document 1 cannot directly determine whether a semiconductor device is quality-critical based on the treatment liquid.

[0005] An object of the present invention is to provide an information processing device, a determination method, and a water treatment system that are capable of detecting defects in semiconductor devices at an early stage. [Means for solving the problem]

[0006] The information processing device of the present invention comprises: an acquisition unit that acquires a liquid quality value of the wastewater of the cleaning liquid used to clean the object; a reading unit that reads out a reference liquid quality value that serves as a reference from the database; a determination unit that determines whether a semiconductor device manufactured in a manufacturing process including a cleaning process of the object is a non-defective product based on the liquid quality value acquired by the acquisition unit and the reference liquid quality value read by the reading unit; and a notification unit that notifies the result of the determination made by the determination unit.

[0007] The determination method of the present invention further comprises: A process of acquiring a liquid quality value of the wastewater of the cleaning liquid used to clean the object; A process of reading out reference liquid quality values ​​stored in a database; a process of determining whether or not the semiconductor device manufactured in the manufacturing process including the cleaning process of the object is a non-defective product based on the acquired liquid quality value and the reference liquid quality value read from the database; The result of the determination is notified.

[0008] In addition, the water treatment system of the present invention includes: a cleaning device that cleans an object using the pure water treated by the water treatment device; a measuring device for measuring the liquid quality value of the wastewater of the cleaning liquid used by the cleaning device to clean the object; a database that stores reference liquid quality values; an information processing device; The information processing device includes: an acquisition unit that acquires the liquid quality value measured by the measuring device; a reading unit that reads the reference fluid quality value from the database; a determination unit that determines whether a semiconductor device manufactured in a manufacturing process including a cleaning process of the object is a non-defective product based on the liquid quality value acquired by the acquisition unit and the reference liquid quality value read by the reading unit; and a notification unit that notifies the result of the determination made by the determination unit. [Effects of the Invention]

[0009] In the present invention, defects in semiconductor devices can be found at an early stage. [Brief explanation of the drawings]

[0010] [Figure 1] 1 is a diagram showing a water treatment system according to a first embodiment. [Figure 2] FIG. 2 is a diagram showing an example of the positioning of the cleaning device shown in FIG. [Figure 3] 2 is a diagram illustrating an example of components included in the information processing device illustrated in FIG. 1. FIG. [Figure 4] FIG. 2 is a diagram showing an example of information stored in a database shown in FIG. [Figure 5] 10 is a flowchart illustrating an example of a method for setting a reference fluid quality value, among the determination methods in the information processing device shown in FIG. [Figure 6] 10 is a flowchart illustrating an example of a method for determining whether a semiconductor device manufactured using a cleaned member as a component is a non-defective product, among determination methods in the information processing apparatus shown in FIG. [Figure 7] 10 is a graph showing an example of reference fluid quality values, which are time-series data, and time-series data of fluid quality values ​​included in fluid quality information. [Figure 8] FIG. 10 is a diagram showing a water treatment system according to a second embodiment. [Figure 9] 9 is a diagram illustrating an example of components included in the information processing device illustrated in FIG. 8. FIG. [Figure 10] 9 is a diagram illustrating an example of input and output of the learning model illustrated in FIG. 8. [Figure 11] 9 is a flowchart illustrating an example of a process in a learning phase of the determination method in the information processing device shown in FIG. 8. [Figure 12] 9 is a flowchart illustrating an example of processing in an inference phase in the determination method in the information processing device shown in FIG. 8. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. (First embodiment)

[0012] Fig. 1 is a diagram showing a water treatment system according to a first embodiment. As shown in Fig. 1, a water treatment system 10 according to this embodiment includes an information processing device 100, a cleaning device 200, a measuring device 300, and a database 400. The information processing device 100 and the database 400 may be directly connected or may be connected via a communication network. Furthermore, the database 400 may be provided within the information processing device 100.

[0013] The cleaning apparatus 200 includes a cleaning section 210 and an exhaust pipe 230. The cleaning apparatus 200 is installed in a semiconductor manufacturing line in a semiconductor manufacturing factory.

[0014] The cleaning unit 210 cleans an object. Here, the object may be an electronic component (finished product) such as a semiconductor device manufactured in a semiconductor device manufacturing process, or a component 220 that constitutes the semiconductor device to be manufactured. In this embodiment, the cleaning unit 210 cleans the component 220, which is a material used in semiconductor element manufacturing (including bare silicon, oxide film, patterned wafers, and semiconductors other than silicon). The cleaning unit 210 performs wet cleaning on the component 220 using a chemical solution or pure water. The cleaning method in the cleaning unit 210 may be a single-tank or multi-tank batch type in which multiple wafers are immersed in a processing tank at the same time and cleaned, or a single-wafer type in which wafers are rotated one by one and cleaned by spraying a processing solution onto them using a nozzle. The cleaning liquids used by the cleaning unit 210 for cleaning include water (including pure water), DHF (Dolute Hydrogen Fluoride), SC1 (a mixed alkaline solution of H2O, ammonia water, and H2O2), SC2 (a mixed acid of H2O, HCl, and H2O2), IPA (Isopropyl Alcohol), TMAH (Tetramethylammonium Hydroxide), etc. The cleaning process performed by the cleaning unit 210 is included in the manufacturing process of semiconductor devices. This manufacturing process includes multiple cleaning processes.

[0015] The discharge pipe 230 discharges the waste liquid used in cleaning the component 220 to the outside of the cleaning device 200. The discharge pipe 230 may separate the waste liquid used in cleaning the component 220 into an organic waste liquid and an inorganic waste liquid, or may separate the waste liquid into an acidic waste liquid and an alkaline waste liquid and then discharge the separated liquid.

[0016] The measuring device 300 measures the liquid quality of the effluent discharged from the discharge pipe 230. The liquid quality measured by the measuring device 300 may be the resistivity, pH, metal concentration, chemical concentration, particulate concentration, ion concentration, total organic carbon (TOC), effluent volume, etc., or any combination thereof. The measuring device 300 may measure the liquid quality of the effluent flowing through the discharge pipe 230 inside the cleaning apparatus 200, or may measure the liquid quality of the effluent flowing through the discharge pipe 230 outside the cleaning apparatus 200. Furthermore, if the discharge pipe 230 is configured to separately discharge acidic effluent and alkaline effluent, the measuring device 300 may be provided with one that measures the liquid quality of the acidic effluent and one that measures the liquid quality of the alkaline effluent. The measuring device 300 transmits liquid quality information including the measured values ​​(liquid quality values) to the information processing device 100. The liquid quality information transmitted by measuring device 300 may include, in addition to the measured liquid quality values, information indicating the type of measured liquid quality (such as the above-mentioned resistivity, pH, metal concentration, chemical concentration, particulate concentration, ion concentration, and drainage volume). Measuring device 300 may transmit to information processing device 100 liquid quality information indicating time-series data of the liquid quality measured in time series from the start time of cleaning of component 220.

[0017] Fig. 2 is a diagram showing an example of the positioning of the cleaning device 200 shown in Fig. 1. In a water treatment system having a pretreatment device 20, a primary pure water production facility 30, and a secondary pure water production facility 40, which is a subsystem, treated water treated in the secondary pure water production facility 40 is supplied to the cleaning device 200 and used to clean a component 220.

[0018] The pretreatment device 20 includes a sand filtration device, an activated carbon device, a softening device, and a decarbonation device. The configuration of the pretreatment device 20 is not limited to this. For example, the configuration of the pretreatment device 20 may be a configuration that does not include some of the devices described above, or a configuration that includes other devices in addition to the above configuration. The sand filtration device uses sand as a filter medium to filter the water to be treated. The activated carbon device uses activated carbon with a porous structure to further filter the treated water from the sand filtration device. The softening device uses an ion exchange resin that removes hardness components from the water to remove hardness components from the treated water from the activated carbon device. The decarbonation device degasses (removes) carbon dioxide gas from the treated water from the softening device using, for example, a decarbonation membrane that allows gas to pass through.

[0019] The primary pure water production facility 30 includes a reverse osmosis (RO) membrane device 31 to which the water to be treated is supplied, an ultraviolet oxidation device 32, an ion exchange device 33, and a membrane degassing device 34. The configuration of the primary pure water production facility 30 is not limited to this configuration. In the primary pure water production facility 30, for example, the order in which the ultraviolet oxidation device 32, the ion exchange device 33, and the membrane degassing device 34 are arranged is not particularly limited. An SB-P (multi-bed regenerative ion exchange device) is provided as the ion exchange device 33. The membrane degassing device 34 removes gas components such as oxygen from the treated water.

[0020] The secondary pure water production equipment 40 includes a primary pure water tank 41, a heat exchanger 42, an ultraviolet oxidation device 43, an ion exchange device 44, a membrane degassing device 45, and an ultrafiltration membrane device 46. Treated water (ultrapure water) produced in the secondary pure water production equipment 40 is supplied to the cleaning equipment 200. The primary pure water tank 41 stores primary pure water produced in the primary pure water production equipment 30. The heat exchanger 42 adjusts the temperature of the primary pure water supplied from the primary pure water tank 41. The ultraviolet oxidation device 43 irradiates the water to be treated, the temperature of which has been adjusted by the heat exchanger 42, with ultraviolet light to decompose organic matter contained in the water. The ion exchange device 44 removes ionic components from the water to be treated treated in the ultraviolet oxidation device 43. The ion exchange device 44 is a non-regenerative cartridge polisher filled with a mixed bed of cation exchange resin and anion exchange resin. The membrane degassing device 45 removes gases such as dissolved oxygen and carbon dioxide contained in the water to be treated that has been treated in the ion exchange device 44. The filtration membrane device 46 is an ultrafiltration membrane device (UF) that removes fine particles contained in the water to be treated that has been treated in the membrane degassing device 45. The configuration of the secondary pure water production equipment 40 is not limited to the above-described configuration. For example, a hydrogen peroxide removal device that decomposes hydrogen peroxide generated in the ultraviolet oxidation device 43 may be provided between the ultraviolet oxidation device 43 and the ion exchange device 44. Impurities removed by the secondary pure water production equipment 40 include metals, ions, TOC (Total Organic Carbon), urea, dissolved oxygen, hydrogen peroxide, fine particles, etc.

[0021] Fig. 3 is a diagram showing an example of components included in the information processing device 100 shown in Fig. 1. As shown in Fig. 3, the information processing device 100 shown in Fig. 1 has an acquisition unit 110, a readout unit 120, a determination unit 130, a notification unit 140, a reference fluid quality value setting unit 150, and a writing unit 160. Fig. 3 shows main elements in this embodiment among the components included in the information processing device 100 shown in Fig. 1.

[0022] Acquiring unit 110 acquires the liquid quality information transmitted by measuring device 300. Acquiring unit 110 outputs the acquired liquid quality information to determining unit 130. The timing at which acquiring unit 110 acquires the liquid quality information is not particularly specified. The measurement unit of the effluent in which the liquid quality value indicated by the liquid quality information acquired by acquiring unit 110 is measured is a predetermined unit in the manufacturing process of component 220. For example, acquiring unit 110 may acquire the liquid quality value of the effluent measured by measuring device 300 for the entire process of one lot, or may acquire each liquid quality value of the effluent measured by measuring device 300 for each of multiple cleaning processes performed in one manufacturing process.

[0023] The reading unit 120 reads out the reference liquid quality values ​​stored in the database 400 from the database 400. For example, the reading unit 120 uses information indicating the type of liquid quality included in the liquid quality information acquired by the acquiring unit 110 as a search key, and reads out the reference liquid quality values ​​associated with the information indicating the type of liquid quality from the database 400. The reading unit 120 outputs the read out reference liquid quality values ​​to the determining unit 130. If the reference liquid quality values ​​stored in the database 400 are time series data with the start time of cleaning the component 220 as the start reference, the reading unit 120 reads out the time series data of the reference liquid quality values ​​as the reference liquid quality values.

[0024] The determination unit 130 determines whether a semiconductor device manufactured using the component 220 cleaned by the cleaning apparatus 200 as a component is a pass-quality product based on the liquid quality value included in the liquid quality information output from the acquisition unit 110 and the reference liquid quality value output from the readout unit 120. Specifically, the determination unit 130 determines that a semiconductor device manufactured using the component 220 cleaned by the cleaning apparatus 200 as a component is a pass-quality product if the difference between the liquid quality value included in the liquid quality information output from the acquisition unit 110 and the reference liquid quality value output from the readout unit 120 is equal to or less than a predetermined threshold. Furthermore, the determination unit 130 determines that a semiconductor device manufactured using the component 220 cleaned by the cleaning apparatus 200 as a component is likely to be defective if the difference between the liquid quality value included in the liquid quality information output from the acquisition unit 110 and the reference liquid quality value output from the readout unit 120 exceeds a predetermined threshold. When the acquiring unit 110 acquires the liquid quality values ​​for each of the multiple cleaning processes, the determining unit 130 may determine that a semiconductor device manufactured using the component 220 cleaned by the cleaning apparatus 200 as a component is likely to be defective if the difference between at least one of the liquid quality values ​​included in the liquid quality information output from the acquiring unit 110 and the reference liquid quality value output from the reading unit 120 exceeds a predetermined threshold. As a result, if the liquid quality value of the wastewater used to clean the component 220 is outside a predetermined reference range, it can be inferred that the cause is some kind of abnormality in the component 220.

[0025] If the liquid quality values ​​included in the liquid quality information output from the acquisition unit 110 are time series data as described above, and the reference liquid quality values ​​stored in the database 400 are time series data with the start time of cleaning of the component 220 as the starting reference, the judgment unit 130 compares the approximation curves (hereinafter referred to as liquid quality curves) of each time series data with the start time of cleaning as the starting reference.

[0026] The threshold value used by the determination unit 130 is a preset value for determining a reference range for determining whether a semiconductor device manufactured using the component 220 as a component is a non-defective product. The larger the threshold value, the wider the range within which a semiconductor device manufactured using the component 220 as a component is determined to be a non-defective product. The smaller the threshold value, the narrower the range within which a semiconductor device manufactured using the component 220 as a component is determined to be a non-defective product. This threshold value is set according to the type of liquid quality being measured. This threshold value may also be set according to the usage status of the cleaning equipment 200, the type and properties of the component 220, and the type of liquid used for cleaning. The threshold value used by the determination unit 130 may be "0." Furthermore, when the above-described liquid quality curve is used, the threshold value is a value (curve) indicating the degree of deviation from the liquid quality curve based on the reference liquid quality value. When the liquid quality curve is used, the threshold value may be a value that changes with the passage of time from the start time of measurement. For example, the threshold value may be set to a larger value closer to the start time of measurement and to a smaller value as time passes from the start time of measurement.

[0027] The determining unit 130 outputs the determination result to the notifying unit 140. For example, if the determination result is that "a semiconductor device manufactured using the member 220 cleaned by the cleaning apparatus 200 as a component is a non-defective product," the determining unit 130 may output a value of "0," and if the determination result is that "a semiconductor device manufactured using the member 220 cleaned by the cleaning apparatus 200 as a component is likely to be defective," the determining unit 130 may output a value of "1." In this case, it goes without saying that the notifying unit 140 must understand the meanings of the values ​​"0" and "1" output from the determining unit 130.

[0028] The notification unit 140 notifies the determination result output from the determination unit 130. The notification by the notification unit 140 may be displayed, output as sound, printed, or transmitted to another device. For example, if the determination result output from the determination unit 130 indicates that a semiconductor device manufactured using the component 220 as a component is likely to be defective, the notification unit 140 may issue a predetermined warning, such as outputting an alarm sound or turning on a lamp. By issuing a warning, the notification unit 140 can quickly notify the user that there is some kind of problem with the component 220.

[0029] The reference liquid-quality value setting unit 150 sets a reference liquid-quality value for each type of liquid quality (such as the resistivity, pH, metal concentration, chemical concentration, particle concentration, ion concentration, and amount of effluent) based on the liquid quality values ​​of the effluent of the cleaning liquid used in the manufacturing process of a semiconductor device that the determining unit 130 determined to be a non-defective product. This inspection data is data obtained by inspecting whether a semiconductor device manufactured using the cleaned component 220 is a non-defective product. The reference liquid-quality value setting unit 150 may set, as the reference liquid-quality value, the liquid quality value that has the best liquid quality value (e.g., a low particle concentration or ion concentration) among the liquid quality values ​​of the effluent used to clean the component 220 constituting a semiconductor device that was determined to be a defective product in the inspection data obtained from a previous inspection. Alternatively, the reference liquid-quality value setting unit 150 may set, as the reference liquid-quality value, the liquid quality value that has the worst liquid quality value (e.g., a high particle concentration or ion concentration) among the liquid quality values ​​of the effluent used to clean the component 220 constituting a semiconductor device that was determined to be a non-defective product in the inspection data obtained from a previous inspection. The reference liquid-quality value setting unit 150 may also calculate and set the reference liquid-quality value using a predetermined calculation algorithm or the deviation of the liquid quality of the wastewater used to clean the component 220 constituting a semiconductor device that was determined to be defective in the inspection data of a previous inspection. The reference liquid-quality value setting unit 150 outputs the set reference liquid-quality value to the writing unit 160.

[0030] The writing unit 160 writes the reference liquid quality values ​​output from the reference liquid quality value setting unit 150 into the database 400. At this time, the writing unit 160 writes the reference liquid quality values ​​for each type of liquid quality into the database 400. By having the writing unit 160 write the reference liquid quality values ​​set by the reference liquid quality value setting unit 150 based on the test data and liquid quality values ​​into the database 400, the reference liquid quality values ​​corresponding to the liquid quality of the wastewater used to wash the component 220 can be used as an appropriate standard for determining whether a semiconductor device manufactured using the component 220 as a constituent element is non-defective.

[0031] The database 400 stores reference liquid quality values ​​for the effluent used to clean the component 220 for each type of liquid quality. FIG. 4 is a diagram showing an example of information stored in the database 400 shown in FIG. 1. As shown in FIG. 4, the database 400 shown in FIG. 1 stores the types of measured liquid qualities associated with reference liquid quality values. The type of measured liquid quality indicates the type of liquid quality for which the reference liquid quality value setting unit 150 has set the reference liquid quality value. Examples of the type of measured liquid quality include the above-mentioned resistivity, pH, metal concentration, chemical concentration, particulate concentration, ion concentration, and effluent volume. The reference liquid quality value is a value set by the reference liquid quality value setting unit 150 for each type of measured liquid quality. These data are written by the writing unit 160.

[0032] The following describes a determination method in the information processing device 100 shown in Fig. 1. Fig. 5 is a flowchart for explaining an example of a method for setting a reference fluid quality value, among the determination methods in the information processing device 100 shown in Fig. 1.

[0033] First, the acquisition unit 110 acquires fluid quality information including the measurement values ​​(fluid quality values) of the effluent measured by the measuring device 300 (step S1). The reference fluid-quality value setting unit 150 stores the fluid quality values ​​acquired by the acquisition unit 110. At this time, the reference fluid-quality value setting unit 150 stores the fluid quality values ​​in a manner that enables identification of which component 220 was cleaned by the effluent for which the fluid quality value was measured. For example, the reference fluid-quality value setting unit 150 may store the fluid quality values ​​acquired by the acquisition unit 110 by adding identification information that can identify the date and time of acquisition, the location, the cleaning device, and the component 220 that was cleaned. Furthermore, the reference fluid-quality value setting unit 150 stores the types of fluid quality values ​​acquired by the acquisition unit 110 in association with the fluid quality values. Furthermore, the reference fluid quality value setting unit 150 may store the fluid quality values ​​acquired by the acquisition unit 110 as reference fluid quality values ​​in a time series (for example, a graph showing the change in fluid quality values ​​over time) during the cleaning process in which the component 220 is cleaned. The reference time for this time series data is the start time of the cleaning process in which the target component 220 is cleaned.

[0034] The acquiring unit 110 then acquires characteristic inspection data for a semiconductor device manufactured using the component 220 as a component (step S2). The characteristic inspection data for the semiconductor device acquired by the acquiring unit 110 is data indicating the results of a characteristic inspection performed on the semiconductor device, and includes information that can identify at least whether the semiconductor device is a non-defective product. At this time, it is necessary to identify which of the liquid property values ​​stored in the reference liquid-property value setting unit 150 corresponds to the characteristic inspection data acquired by the acquiring unit 110. For example, when the acquiring unit 110 acquires a liquid property value, the reference liquid-property value setting unit 150 may store identification information for the component 220 that was washed with the wastewater of that liquid property value. When the acquiring unit 110 acquires characteristic inspection data, the acquiring unit 110 may acquire identification information for the component 220 that constitutes the semiconductor device that is the subject of inspection for the characteristic inspection data, and the reference liquid-property value setting unit 150 may compare the identification information with each other.

[0035] Once the correspondence between the liquid property value and the characteristic inspection data has been recognized, a result is obtained indicating whether or not a semiconductor device manufactured using the component 220 to be cleaned, for which the obtained liquid property value has been obtained, is a pass / fail product. In other words, the liquid property value is associated with information indicating whether or not the semiconductor device is a pass / fail product. Based on this association, the reference liquid property value setting unit 150 sets a reference liquid property value (step S3). Then, the writing unit 160 writes the reference liquid property value set by the reference liquid property value setting unit 150 to the database 400 (step S4).

[0036] Fig. 6 is a flowchart illustrating an example of a method for determining whether a semiconductor device manufactured using a cleaned member 220 as a component is non-defective, among the determination methods in the information processing device 100 shown in Fig. 1. The process described below is performed after the reference liquid quality value is set in the process described using the flowchart shown in Fig. 5.

[0037] First, the acquisition unit 110 acquires liquid quality information including the measurement values ​​(liquid quality values) of the effluent measured by the measuring device 300 (Step S11). At this time, the acquisition unit 110 may acquire liquid quality information including time-series data of the measurement values ​​(liquid quality values) of the effluent measured by the measuring device 300. The reference time of the time-series data is the start time of the cleaning process for cleaning the target component 220. Then, the reading unit 120 uses the type of measured liquid quality included in the liquid quality information as a search key to search and read out a reference liquid quality value stored in association with the type of measured liquid quality from the database 400 (Step S12). Next, the determination unit 130 compares the liquid quality value included in the liquid quality information acquired by the acquisition unit 110 with the reference liquid quality value read out by the reading unit 120 (Step S13). The determination unit 130 determines whether the difference between the liquid quality value and the reference liquid quality value is equal to or less than a threshold (Step S14).

[0038] If the difference between the liquid property value and the reference liquid property value is equal to or less than the threshold, the determination unit 130 determines that the semiconductor device manufactured using the component 220 is a non-defective product (step S15). On the other hand, if the difference between the liquid property value and the reference liquid property value exceeds the threshold, the determination unit 130 determines that the semiconductor device manufactured using the component 220 is likely to be defective (step S16). Then, the notification unit 140 notifies the determination result of the determination unit 130 (step S17).

[0039] If the reference liquid quality values ​​stored in the database 400 and the liquid quality values ​​included in the liquid quality information acquired by the acquisition unit 110 are time-series data as described above, in step S13, the determination unit 130 compares the time-series data of the reference liquid quality values ​​with the time-series data of the liquid quality values ​​included in the liquid quality information acquired by the acquisition unit 110. FIG. 7 is a graph showing an example of the time-series data of the reference liquid quality values ​​and the time-series data of the liquid quality values ​​included in the liquid quality information acquired by the acquisition unit 110. FIG. 7 also shows two time-series data of liquid quality values ​​that the determination unit 130 has determined to be abnormal (possibly defective) as the liquid quality values ​​included in the liquid quality information acquired by the acquisition unit 110. In the example shown in FIG. 7, the liquid quality values ​​that fall within the reference liquid quality values ​​decrease (the liquid quality improves) as the cleaning time elapses during the cleaning process. Also, in FIG. 7, the range in which the difference from the reference liquid quality value is equal to or less than a threshold is displayed in gray. Note that the vertical axis of the graph shown in FIG. 7 indicates the value of an arbitrary type of liquid quality value and does not indicate the actual measured value. The reference time for this time series data is the start time of the cleaning process for cleaning the target component 220. In abnormal result 1 shown in FIG. 7, the liquid quality value is outside the threshold value at around 20 seconds. In abnormal result 2, the liquid quality value is outside the threshold value between 0 and 7 seconds. In such cases, the determination unit 130 determines that the semiconductor device manufactured using the component 220 in question as a component is likely to be defective.

[0040] It is also possible to set a plurality of the above thresholds. In this case, for example, "threshold A", "threshold B" and "threshold C" can be set as follows: Threshold A<Threshold B<Threshold C The determination unit 130 may set a magnitude relationship such that the difference between the liquid quality value included in the liquid quality information acquired by the acquisition unit 110 and the reference liquid quality value read by the readout unit 120 is compared with each of thresholds A to C. If the difference is equal to or less than threshold A, the determination unit 130 may determine that the semiconductor device manufactured using the component 220 is a pass / fail product. If the difference is greater than threshold A and equal to or less than threshold B, the determination unit 130 may determine that the semiconductor device manufactured using the component 220 is a pass / fail product with a 30% probability. If the difference is greater than threshold B and equal to or less than threshold C, the determination unit 130 may determine that the semiconductor device manufactured using the component 220 is a pass / fail product with a 70% probability. If the difference is greater than threshold C, the determination unit 130 may determine that the semiconductor device manufactured using the component 220 is a pass / fail product with a 90% probability. In this manner, multiple thresholds having different magnitudes may be provided, and the determination unit 130 may determine the probability that the semiconductor device is a pass / fail product based on the results of comparing the difference with each threshold.

[0041] As described above, in this embodiment, whether a semiconductor device manufactured using the component 220 as a component is a pass / fail product is determined based on the liquid quality value of the effluent used to clean the component 220. A preset reference liquid quality value is used for the determination. The reference liquid quality value is set using the liquid quality value of the effluent used to clean the component 220 constituting the semiconductor device and the results of inspecting the characteristics of the semiconductor device. If the difference between the liquid quality value of the effluent used to clean the component 220 and the reference liquid quality value is equal to or less than a threshold, the semiconductor device manufactured using the component 220 as a component is determined to be a pass / fail product. If the difference between the liquid quality value and the reference liquid quality value exceeds the threshold, the semiconductor device manufactured using the component 220 as a component is determined to be potentially defective. This allows for early detection of defects in semiconductor devices without waiting for the results of a characteristic inspection performed on the semiconductor device after cleaning. (Second embodiment)

[0042] FIG. 8 is a diagram showing a water treatment system according to a second embodiment. As shown in FIG. 8, the water treatment system 11 according to this embodiment includes an information processing device 101, a cleaning device 200, a measuring device 300, and a learning model 500. The cleaning device 200 and the measuring device 300 are the same as those in the first embodiment. The information processing device 101 and the learning model 500 may be directly connected or may be connected via a communication network. Furthermore, the learning model 500 may be provided within the information processing device 101.

[0043] Fig. 9 is a diagram showing an example of components included in the information processing device 101 shown in Fig. 8. As shown in Fig. 9, the information processing device 101 shown in Fig. 8 has an acquisition unit 110, a determination unit 131, a notification unit 140, and a learning model generation unit 171. The acquisition unit 110 and the notification unit 140 are the same as those in the first embodiment. Note that Fig. 9 shows main elements in this embodiment among the components included in the information processing device 101 shown in Fig. 8.

[0044] The learning model generation unit 171 stores, as training data, liquid property values ​​corresponding to inspection data obtained by inspecting whether a semiconductor device manufactured using the cleaned member 220 as a component is non-defective, and performs machine learning on the liquid property values ​​to generate a learning model 172. The association between the inspection data and the liquid property values ​​is the same as in the first embodiment. The learning model generation unit 171 is used in the so-called learning phase (training phase) of machine learning.

[0045] The learning model 500 is a learning model generated by the learning model generation unit 171. The learning model 500 outputs a judgment result indicating whether a semiconductor device manufactured using the component 220 cleaned with wastewater, for which a liquid quality value has been acquired using the training data generated by the learning model generation unit 171, is a pass / fail product. The learning model 500 may have, for example, a neural network structure in which multiple neurons are interconnected. A neuron is an element that performs a predetermined calculation on multiple inputs and outputs a single value as the calculation result. The learning model 500 is stored in a storage device (not shown). The learning method used in the learning model 500 may be a general method for generating a learning model. FIG. 10 is a diagram illustrating an example of input / output of the learning model 500 shown in FIG. 8. As shown in FIG. 10, the learning model 500 shown in FIG. 8 is a learning model that, when a liquid quality value is input, outputs a judgment result indicating whether a semiconductor device manufactured using the component 220 is a pass / fail product. A learning model 500 is provided for each type of measured liquid quality and each type of cleaning liquid. Furthermore, the learning model 500 may be provided for each usage situation of the cleaning device 200 and each type and property of the component 220 .

[0046] The determination unit 131 inputs the liquid quality values ​​included in the liquid quality information acquired by the acquisition unit 110 into the learning model 500, and acquires a determination result indicating whether a semiconductor device manufactured using the member 220, which was cleaned with the wastewater from which the liquid quality values ​​were acquired, as a component, is a non-defective product from the learning model 500. The determination unit 131 is used in the so-called inference phase of machine learning.

[0047] The following describes a determination method in the information processing device 101 shown in Fig. 8. Fig. 11 is a flowchart for explaining an example of processing in the learning phase of the determination method in the information processing device 101 shown in Fig. 8.

[0048] First, the acquisition unit 110 acquires fluid quality information including the measurement values ​​(fluid quality values) of the effluent measured by the measuring device 300 (step S21). The learning model generation unit 171 stores the fluid quality values ​​acquired by the acquisition unit 110. At this time, the learning model generation unit 171 stores the fluid quality values ​​in a manner that enables identification of which component 220 the effluent for which the fluid quality value was measured was cleaned. For example, the learning model generation unit 171 may store the fluid quality values ​​acquired by the acquisition unit 110 by adding identification information that can identify the date and time of acquisition, the location, the cleaning device, and the component 220 that was cleaned. Furthermore, the learning model generation unit 171 stores the types of fluid quality values ​​acquired by the acquisition unit 110 in association with the fluid quality values.

[0049] Thereafter, the acquiring unit 110 acquires characteristic inspection data of the semiconductor device (step S22). The characteristic inspection data of the semiconductor device acquired by the acquiring unit 110 is data indicating the results of a characteristic inspection of the semiconductor device, and includes at least information that can identify whether the semiconductor device is a non-defective product. At this time, it is necessary to identify which of the liquid property values ​​stored in the learning model generation unit 171 corresponds to the characteristic inspection data acquired by the acquiring unit 110. For example, when the acquiring unit 110 acquires a liquid property value, the learning model generation unit 171 may store identification information of the components 220 that constitute the semiconductor device that was cleaned with the wastewater of the liquid property value. When the acquiring unit 110 acquires characteristic inspection data, the acquiring unit 110 may acquire identification information of the components 220 that constitute the semiconductor device that is the subject of inspection for the characteristic inspection data, and the learning model generation unit 171 may compare the identification information with each other.

[0050] Once the correspondence between the liquid property value and the characteristic inspection data has been recognized, a result indicating whether a semiconductor device manufactured using the component 220 to be cleaned, which corresponds to the obtained liquid property value, is a pass / fail product is obtained. In other words, the liquid property value is associated with information indicating whether a semiconductor device manufactured using the component 220 is a pass / fail product. The learning model generation unit 171 generates a learning model by having the learning model 500 learn this correspondence (step S23). At this time, the learning model generation unit 171 causes the learning model 500 corresponding to the type of liquid property value to perform learning. If a learning model 500 is not provided for each type of liquid property value and the learning model 500 is a learning model that determines whether a semiconductor device is a pass / fail product based on the liquid property value and the type of liquid property value, the learning model generation unit 171 causes the learning model 500 to learn the liquid property value, the type of liquid property value, and information indicating whether the semiconductor device is a pass / fail product.

[0051] Fig. 12 is a flowchart for explaining an example of the processing of the inference phase in the determination method in the information processing device 101 shown in Fig. 8. The processing described below is performed after the learning phase described using the flowchart shown in Fig. 11 has been performed.

[0052] First, the acquisition unit 110 acquires fluid quality information including the measurement values ​​(fluid quality values) of the effluent measured by the measuring device 300 (step S31). Next, the determination unit 131 inputs the fluid quality values ​​included in the fluid quality information into the learning model 500 corresponding to the type of fluid quality value (step S32). If a learning model 500 is not provided for each type of fluid quality value, the determination unit 131 inputs the type of fluid quality value along with the fluid quality value into the learning model 500. When the learning model 500 to which the fluid quality values ​​have been input outputs a result, the determination unit 131 acquires the judgment result output by the learning model 500 (step S33). This judgment result is information indicating whether the semiconductor device is a non-defective product. Next, the notification unit 140 notifies the determination result acquired by the determination unit 131 (step S34).

[0053] In addition, using the judgment result notified in step S34 and the results of a characteristic inspection of the semiconductor device that is conducted subsequently, the learning model generation unit 171 may re-learn the learning model 500 to improve the judgment accuracy.

[0054] Furthermore, a learning model may be provided that uses feedback based on the results of the characteristic inspection of the semiconductor device and the measured liquid quality values ​​to learn what kind of defect is detected in a semiconductor device manufactured using the part 220 as a component when what kind of liquid quality value is measured, and a system may be constructed that infers the nature of the defect in the semiconductor device based on the measured liquid quality values.

[0055] As described above, in this embodiment, a determination is made as to whether a semiconductor device is a pass / fail product based on the liquid quality values ​​of the waste liquid used to clean the components 220 constituting the semiconductor device. The learning model 500 is used for the determination. When the liquid quality values ​​of the waste liquid used to clean the components 220 constituting the semiconductor device are input, the learning model 500 outputs a determination result indicating whether the semiconductor device is a pass / fail product. Therefore, defects in semiconductor devices can be detected early without waiting for the results of a characteristic inspection performed on the semiconductor device after cleaning. Furthermore, the use of the learning model 500 can handle a large number of liquid quality value patterns, thereby obtaining useful determination results.

[0056] Although the above description has been given by allocating each function (process) to each component, this allocation is not limited to the above. Furthermore, the configuration of the components is also not limited to the above-described embodiments, which are merely examples. Furthermore, the above-described embodiments may be combined in any combination.

[0057] The processes performed by each of the information processing devices 100 and 101 described above may be performed by logic circuits manufactured for each purpose. Alternatively, a computer program (hereinafter referred to as a program) describing the process contents as procedures may be recorded on a recording medium readable by each of the information processing devices 100 and 101, and the program recorded on the recording medium may be read and executed by each of the information processing devices 100 and 101. Examples of recording media readable by each of the information processing devices 100 and 101 include removable recording media such as floppy (registered trademark) disks, magneto-optical disks, DVDs (Digital Versatile Discs), CDs (Compact Discs), Blu-ray (registered trademark) Discs, USB (Universal Serial Bus) memories, and SD cards, as well as memories such as ROMs (Read Only Memory), RAMs (Random Access Memory), and HDDs (Hard Disc Drives) built into each of the information processing devices 100 to 102. The program recorded on this recording medium is read by a CPU (not shown) provided in each of the information processing devices 100 and 101, and the same processing as described above is performed under the control of the CPU. Here, the CPU operates as a computer that executes the program read from the recording medium on which the program is recorded.

[0058] Some of the above embodiments can be described as, but are not limited to, the following supplementary notes. (Supplementary Note 1) An acquisition unit that acquires a liquid quality value of the drained cleaning liquid used to clean the object; a learning model generation unit that stores, as training data, the liquid property values ​​corresponding to inspection data obtained by inspecting whether or not a semiconductor device manufactured in a manufacturing process including a process for cleaning the object after cleaning is a non-defective product, performs machine learning on the liquid property values, and generates a learning model that uses the liquid property values ​​as an input and outputs a determination result indicating whether or not the semiconductor device is a non-defective product; a determination unit that inputs the liquid quality values ​​acquired by the acquisition unit into the learning model and acquires from the learning model a determination result indicating whether the semiconductor device cleaned with the wastewater from which the liquid quality values ​​were acquired is a non-defective product; and a notification unit that notifies the determination result acquired by the determination unit. [Explanation of symbols]

[0059] 10,11 Water treatment systems 20 Pretreatment device 30 Primary pure water production equipment 31 Reverse osmosis membrane equipment 32,43 Ultraviolet oxidation device 33,44 Ion exchange device 34,45 Membrane degassing device 40 Secondary pure water production equipment 41 Primary pure water tank 42 Heat exchanger 46 Membrane filtration equipment 100,101 Information processing equipment 110 Acquisition Department 120 Readout section 130,131 Judgment part 140 Notification Department 150 Reference liquid quality value setting section 160 Writing section 171 Learning model generation unit 200 Cleaning Equipment 210 Cleaning section 220 parts 230 Discharge pipe 300 measuring instruments 400 databases 500 Learning Models

Claims

1. an acquisition unit that acquires a liquid quality value of the wastewater of the cleaning liquid used to clean the object; a reading unit that reads out a reference liquid quality value that serves as a reference from the database; a determination unit that determines whether a semiconductor device manufactured in a manufacturing process including a cleaning process of the object is a non-defective product based on the liquid quality value acquired by the acquisition unit and the reference liquid quality value read by the reading unit; and a notification unit that notifies the result of the determination made by the determination unit.

2. 2. The information processing device according to claim 1, The judgment unit judges that the semiconductor device is a good product if the difference between the liquid quality value and the reference liquid quality value is less than a predetermined threshold, and judges that the semiconductor device manufactured in the manufacturing process is likely to be defective if the difference between the liquid quality value and the reference liquid quality value exceeds the threshold.

3. 3. The information processing device according to claim 1, a reference liquid quality value setting unit that sets the reference liquid quality value based on a liquid quality value of a waste cleaning liquid used in a manufacturing process of a semiconductor device that has been determined to be a non-defective by the determination unit; a writing unit that writes the reference fluid quality value set by the reference fluid quality value setting unit into the database.

4. 3. The information processing device according to claim 1, the manufacturing process includes multiple cleaning steps; the acquiring unit acquires a liquid quality value of the wastewater measured by a measuring device in each of the plurality of cleaning processes, The determination unit determines that the semiconductor device is likely to be defective if the difference between at least one of the liquid quality values ​​of the wastewater measured by the measuring device in each of the multiple cleaning processes and the reference liquid quality value exceeds a threshold value.

5. 3. The information processing device according to claim 1, The information processing apparatus, wherein the wastewater for which the liquid quality value acquired by the acquisition unit is measured is wastewater used in cleaning wafers or patterned wafers used in the semiconductor devices.

6. 3. The information processing device according to claim 2, The notification unit issues a predetermined warning when the determination unit determines that the semiconductor device is likely to be defective.

7. 3. The information processing device according to claim 1, the acquisition unit acquires time-series data of the liquid quality value based on a start time of the cleaning; the reading unit reads the time series data of the reference fluid quality value from the database; The determination unit compares the time series data of the liquid quality value acquired by the acquisition unit with the time series data of the reference liquid quality value read from the database by the read unit, and determines whether or not a semiconductor device manufactured in a manufacturing process that includes a cleaning process of the object is a non-defective product based on the result of the comparison.

8. A process of acquiring a liquid quality value of the wastewater of the cleaning liquid used to clean the object; A process of reading out reference liquid quality values ​​stored in a database; a process of determining whether or not the semiconductor device manufactured in the manufacturing process including the cleaning process of the object is a non-defective product based on the acquired liquid quality value and the reference liquid quality value read from the database; and notifying the result of the determination.

9. a cleaning device that cleans an object using the pure water treated by the water treatment device; a measuring device for measuring the liquid quality value of the wastewater of the cleaning liquid used by the cleaning device to clean the object; a database that stores reference liquid quality values; an information processing device; The information processing device includes: an acquisition unit that acquires the liquid quality value measured by the measuring device; a reading unit that reads the reference fluid quality value from the database; a determination unit that determines whether a semiconductor device manufactured in a manufacturing process including a cleaning process of the object is a non-defective product based on the liquid quality value acquired by the acquisition unit and the reference liquid quality value read by the reading unit; a notification unit that notifies the result of the determination made by the determination unit.

Images