Abnormality detection method and substrate treatment device

Abstract

An abnormality detection method for a hot plate for treating a substrate, the method comprising: a measurement step for measuring and storing changes in temperature in the hot plate during substrate treatment; a temperature integration step for integrating the changes in temperature during a plurality of substrate treatments stored in the measurement step; a normal population creation step for creating a normal population that is a normal integration result of a prescribed number of substrates integrated in the temperature integration step; and an abnormality detection step for creating an inspection target population that is an integration result of a predetermined number of inspection target substrates integrated in the temperature integration step, and detecting an abnormality related to the hot plate by comparing the inspection target population with the normal population.

Description

Abnormality detection method and substrate processing apparatus

[0001] The present disclosure relates to an abnormality detection method and a substrate processing apparatus.

[0002] Patent Document 1 discloses a thermal processing unit including a hot plate for placing a wafer and applying heat to the wafer, a heater for heating the hot plate, multiple temperature sensors corresponding to multiple channels of the hot plate for measuring the temperature of the hot plate, and a controller. The controller calculates a temperature shift amount, which is the difference between the temperature indicated by the temperature sensor and the ideal temperature according to the heater setting, for each of the multiple channels, and determines whether the temperature shift amount is within a predetermined bandwidth. Furthermore, if there is a channel whose temperature shift amount is not within the bandwidth, the controller identifies that channel as an abnormal region.

[0003] Patent Publication No. 2019-65126

[0004] The present disclosure aims to appropriately detect defects in a hot plate.

[0005] The anomaly detection method according to the present disclosure is an anomaly detection method relating to a hot plate for processing substrates, and includes a measurement process for measuring and storing temperature changes on the hot plate during substrate processing, a temperature accumulation process for accumulating the temperature changes during processing of multiple substrates stored in the measurement process, a normal population creation process for creating a normal population which is a normal accumulation result for a predetermined number of substrates accumulated in the temperature accumulation process, and an anomaly detection process for detecting an anomaly relating to the hot plate by creating a test subject population which is an accumulation result for a predetermined number of substrates accumulated in the temperature accumulation process, and comparing the test subject population with the normal population.

[0006] According to the present disclosure, defects in the heating plate can be appropriately detected.

[0007] 1 is a plan view schematically showing an outline of the configuration of a wafer processing system according to an embodiment of the present invention; FIG. 2 is a front view schematically showing an outline of the configuration of a wafer processing system according to an embodiment of the present invention; FIG. 3 is a schematic vertical sectional view showing an example of a heat processing unit; FIG. 4 is a diagram explaining that an integrated value (average value) monotonically increases according to a set temperature; FIG. 5 is a diagram explaining control related to a warped wafer; FIG. 6 is a diagram explaining overshoot monitoring; FIG. 7 is a diagram explaining a sampling example; FIG. 8 is a diagram explaining a sampling example; FIG. 9 is a diagram explaining Welch's t-test; FIG. 10 is a diagram explaining an anomaly detection process; FIG. 11 is a diagram explaining an anomaly detection process; FIG. 12 is a diagram explaining a health check; FIG. 13 is a flowchart showing the overall flow of anomaly detection; FIG. 14 is a flowchart showing details of normal population creation; FIG. 15 is a flowchart showing details of anomaly detection; FIG. 16 is a graph explaining an anomaly detection method according to a first modified example; FIG. 17 is a graph explaining an anomaly detection method according to a second modified example; FIG. 18 is a graph explaining an anomaly detection method according to a third modified example; FIG. 19 is a diagram explaining an anomaly detection method according to a fourth modified example.

[0008] Hereinafter, a wafer processing system as a substrate processing apparatus according to the present embodiment will be described with reference to the drawings. In this specification, elements having substantially the same functional configurations are designated by the same reference numerals, and redundant description will be omitted.

[0009] <Wafer Processing System> First, the configuration of a wafer processing system according to this embodiment will be described. Figures 1 and 2 are a plan view and a front view, respectively, that schematically show the overall configuration of a wafer processing system 1. In this embodiment, the wafer processing system 1 will be described as an example of a photolithography processing system that performs a resist film forming process and a development process on a wafer W.

[0010] 1, the wafer processing system 1 includes a cassette station 2 into which a cassette C containing a plurality of wafers W is loaded and unloaded, and a processing station 3 equipped with a plurality of various processing devices that perform predetermined processing on the wafers W. The wafer processing system 1 has a configuration in which the cassette station 2, the processing station 3, and an interface station 4 that transfers the wafers W between them and an exposure device (not shown) adjacent to the opposite side of the processing station 3 are integrally connected. Note that, although two processing stations 3 are installed between the cassette station 2 and the interface station 4 as shown in FIG. 1, one processing station 3 or three or more processing stations may be installed.

[0011] The cassette station 2 is provided with a plurality of cassette mounting tables 21 and wafer transfer devices 22 and 23. The cassette station 2 transfers wafers between the cassette C mounted on the mounting table 12 and the processing station 3 using the wafer transfer device 22 or 23. Each of the wafer transfer devices 22 and 23 is provided with a drive mechanism having movement paths in various directions, such as horizontal directions (X direction and Y direction), vertical directions (Z direction), and around a vertical axis (θ direction), as needed. Alternatively, a drive mechanism having movement paths in all directions may be provided. At least one of the wafer transfer devices 22 and 23 is capable of transferring wafers between the cassette C and the processing station 3. Transferring wafers between the processing station 3 and a third block G3 includes a transfer device accessible by a wafer transfer device 33 in the processing station 3 (described later). The third block G3 may include a plurality of transfer devices (not shown) arranged vertically.

[0012] An inspection device (not shown) for inspecting the wafer W may be provided at a position accessible to either the wafer transfer device 22 or 23 .

[0013] The processing station 3 includes multiple blocks, e.g., three blocks G1, G2, and G4 (first, second, and fourth blocks). As shown in FIG. 2, multiple layers 31 each including the first and second blocks G1 and G2 are stacked vertically. For example, the first block G1 is provided on the front side of the processing station 3 (the negative X-direction side in FIG. 1), and the second block G2 is provided on the rear side of the processing station 3 (the positive X-direction side in FIG. 1). The fourth block G4 is provided on the interface station 5 side of the processing station 3 (the positive Y-direction side in FIG. 1) or at a connection point with another adjacent processing station 3. The fourth block G4 may include multiple transfer devices arranged vertically. The aforementioned third block G3 may also be provided within the processing station 3.

[0014] The first block G1 includes a plurality of liquid processing devices, such as a patterning film forming device and a development processing device (both not shown). The patterning film forming device may include, for example, a resist film forming device and an anti-reflection film forming device.

[0015] For example, a plurality of processing devices may be arranged in a horizontal line, and the number, arrangement, and type of these processing devices may be selected arbitrarily.

[0016] In these patterning film forming apparatuses and developing treatment apparatuses, for example, a predetermined processing liquid or a predetermined gas is supplied onto the wafer W. In this manner, the patterning film forming apparatus forms a resist film used as a mask when forming a pattern on an underlying film, or forms an anti-reflection film for efficiently performing a light irradiation process, such as an exposure process. On the other hand, in the developing treatment apparatus, a portion of the exposed resist film is removed to form the uneven shape that serves as the mask.

[0017] For example, the second block G2 is provided with vertically and horizontally aligned thermal processing units U2 that perform thermal processing such as heating and cooling of the wafer W. The second block G2 also is provided with vertically (Z direction in FIG. 2 ) and horizontally aligned hydrophobization processing devices that perform hydrophobization processing to improve the fixation of the resist liquid to the wafer W, and peripheral exposure devices that expose the peripheral portion of the wafer W, neither of which are shown. The number and arrangement of these thermal processing units U2, hydrophobization processing devices, and peripheral exposure devices can also be selected as desired.

[0018] 1, a wafer transfer area 32 is formed in an area sandwiched between the first block G1 and the second block G2 in a plan view. In the wafer transfer area 32, for example, a wafer transfer device 33 is disposed.

[0019] The wafer transfer device 33 has a transfer arm 70a that is movable in, for example, the Y direction, the front-rear direction, the θ direction, and the up-down direction. The wafer transfer device 70 moves within the wafer transfer region 32 and can transfer the wafer W to predetermined devices in the surrounding first block G1, second block G2, third block G3, and fourth block G4. The wafer transfer device 33 provided in the processing station 3 located on the interface station 5 side can transfer the wafer W to predetermined devices in the fifth block G5 (described below) in addition to the first, second, and fourth blocks G1, G2, and G3.

[0020] 2, for example, a plurality of wafer transfer devices 33 are arranged one above the other. One wafer transfer device 33 can transfer wafers W to a predetermined device located at the height of the upper layers 31 among the plurality of layers 31 stacked vertically. Another wafer transfer device 33 can transfer wafers W to a predetermined device located at the height of the plurality of layers 31 located below the above layers 31. A plurality of wafer transfer regions 32 are provided to enable such transfer of wafers W. Note that the number of wafer transfer devices 33 and the number of layers 31 corresponding to one wafer transfer device 33 can be selected arbitrarily, such as by providing a wafer transfer device 33 for each layer 31.

[0021] The wafer transfer area 32, the first block G1, or the second block G2 may also include a shuttle transfer device (not shown). The shuttle transfer device linearly transfers wafers W between a space adjacent to one side of the processing station 3 and another space adjacent to the opposite side.

[0022] The interface station 4 is provided with a fifth block G5 equipped with a plurality of transfer devices, and wafer transfer devices 41 and 42. The interface station 4 uses the wafer transfer device 41 or 42 to transfer the wafer W between the fifth block G5, where the wafer W is transferred by the wafer transfer device 33, and the exposure device. To this end, the wafer transfer devices 41 and 42 are each provided with a drive mechanism having movement paths in various directions, such as the horizontal direction (X direction, Y direction), the vertical direction (Z direction), and around the vertical axis (θ direction), as needed, or may be provided with a drive mechanism having movement paths in all directions. At least one of the wafer transfer devices 41 and 42 can support the wafer W and transfer the wafer W between the transfer device in the fifth block G5 and the exposure device.

[0023] A cleaning device for cleaning the surface of the wafer W and the aforementioned peripheral exposure device may be provided in the interface station 4 at a position accessible to either of the wafer transfer devices 41 and 42 .

[0024] The inspection device may also be provided in the processing station 3 and the interface station 4 at a position accessible by any of the transport arms (33, 41, 42 in FIG. 1 or 2) provided inside each station.

[0025] The wafer processing system 1 described above is provided with a controller 100 (controller). The controller 100 is, for example, a computer and has a program storage unit (not shown). The program storage unit stores a program for controlling the processing of wafers W in the wafer processing system 1. The program storage unit also stores a program for controlling the operation of drive systems such as the various processing devices and transport devices described above to perform wafer processing in the wafer processing system 1. The program may be recorded on a computer-readable storage medium H and installed into the controller 100 from the storage medium H. The storage medium H may include a ROM, RAM, or hard disk, but the structure and type are not limited, and the storage medium H may be temporary or non-temporary. The controller 100 may include a unit for storing, reading, and executing the program for performing wafer processing and for performing related communications. Each unit may be located inside or outside the wafer processing system 1. The controller 100 may be one or more circuits, or may be provided as an integrated unit or as separate units.

[0026] <Operation of Wafer Processing System> The wafer processing system 1 is configured as described above. Next, an example of wafer processing performed using the wafer processing system 1 configured as described above will be described.

[0027] First, a cassette C containing a plurality of wafers W is carried into the cassette station 2 of the wafer processing system 1 and placed on the cassette mounting table 21. Next, each wafer W in the cassette C is sequentially removed by the wafer transfer device 22 or 23 and transferred to the delivery device in the third block G3.

[0028] The wafer W transferred to the transfer device in the third block G3 is supported by the wafer transfer device 33 and transferred to the hydrophobization treatment device provided in the second block G2, where hydrophobization treatment is performed. The wafer W is then transferred by the wafer transfer device 33 to the resist film forming device, where a resist film is formed on the wafer W, and then transferred to the thermal processing unit U2, where a pre-bake treatment is performed, before being transferred to the transfer device in the fifth block G5. When there are multiple processing stations 3 as shown in FIGS. 1 and 2 , the wafer W is temporarily placed in the transfer device in the fourth block G4 before being transferred to the transfer device in the fifth block G5, and then transferred between the multiple wafer transfer devices 33. If necessary, the wafer W may be transferred by the wafer transfer device 33 to a peripheral exposure device, where the peripheral edge of the wafer is exposed.

[0029] The wafer W transferred to the delivery device in the fifth block G5 is transferred to the exposure device by wafer transfer devices 41 and 42, and is exposed to a predetermined pattern. The wafer W may be cleaned in a cleaning device before the exposure process.

[0030] The exposed wafer W is transferred to the delivery device in the fifth block G5 by the wafer transfer devices 41 and 42. Thereafter, the wafer W is transferred to the thermal processing unit U2 by the wafer transfer device 33, where it is subjected to post-exposure baking.

[0031] The wafer W that has been subjected to post-exposure baking is transferred by the wafer transfer device 33 to a developing treatment device and developed. After development is completed, the wafer W is transferred by the wafer transfer device 33 to the heat treatment unit U2 and subjected to post-baking.

[0032] Thereafter, the wafer W is transferred by the wafer transfer device 33 to the delivery device in the third block G3, and then transferred by the wafer transfer device 22 or 23 in the cassette station 2 to the cassette C on the predetermined cassette mounting table 21. In this way, a series of photolithography steps is completed.

[0033] It should be noted that the wafer processing system of the present disclosure is not limited to the configuration and operation described above. For example, in the above embodiment, the wafer processing system is directly connected to the exposure apparatus, and wafers W are transferred between the interface station 4 and the exposure apparatus. However, the wafer processing system does not have to be directly connected to the exposure apparatus. In that case, for example, the wafer W is transferred from the cassette station 2 to the processing station 3, where it undergoes the necessary processing, and then transferred back to the cassette station 2 for removal from the system. Furthermore, unnecessary processing equipment listed above may not be provided in the wafer processing system, or processing may not be performed in that equipment.

[0034] <Heat-Treatment Unit> Next, the heat-treating unit U2 will be described in detail. As shown in Fig. 3, the heat-treating unit U2 includes a housing 90, a heating mechanism 130, a temperature adjustment mechanism 150, and a controller 100 (controller).

[0035] The housing 90 is a processing vessel that houses the heating mechanism 130 and the temperature adjustment mechanism 150. A loading port 91 for the wafer W is opened in a side wall of the housing 90. A floor plate 92 is provided inside the housing 90 to divide the interior of the housing 90 into an upper region, which is a region where the wafer W moves, and a lower region.

[0036] The heating mechanism 130 is configured to heat the wafer W. The heating mechanism 130 includes a support table 131, a top plate 132, a lifting mechanism 133, a hot plate 134, support pins 135, a lifting mechanism 136, an exhaust duct 137, a heater 138 (temperature controller), and a temperature sensor 139.

[0037] The support base 131 is a cylindrical member with a recess formed in the center. The support base 131 supports a hot plate 134. The top plate 132 is a disk-shaped member with a diameter approximately the same as that of the support base 131. The top plate 132 is supported, for example, on the ceiling of the housing 90, and faces the support base 131 across a gap. An exhaust duct 137 is connected to the top of the top plate 132. The exhaust duct 137 exhausts air from the chamber.

[0038] The lifting mechanism 133 is configured to lift and lower the top plate portion 132 under the control of the controller 100. When the top plate portion 132 is raised by the lifting mechanism 133, a chamber, which is a space where the heat treatment of the wafer W is performed, is opened, and when the top plate portion 132 is lowered, the chamber is closed.

[0039] The hot plate 134 is, for example, a circular flat plate, and is fitted into a recess in the support base 131. The hot plate 134 mounts a wafer W and applies heat to the wafer W. The hot plate 134 is heated by a heater 138. The hot plate 134 may be heated by the heater 138 for each of a plurality of channels (regions). A temperature sensor 139 configured to measure the temperature of the hot plate 134 is provided inside the hot plate 134. In this embodiment, a single temperature sensor 139 is provided for the hot plate 134, but multiple temperature sensors 139 may be provided for the hot plate 134 (for example, for each of the above-mentioned plurality of regions).

[0040] The heater 138 is a temperature regulator that heats the hot plate 134. The heater 138 is composed of, for example, a resistance heating element. The heater 138 may be configured to heat multiple channels of the hot plate 134 according to a command temperature set by the controller 100. That is, a command temperature may be set for each of the multiple channels in the heater 138. The command temperature for each channel may be individually changeable by the controller 100. The heater 138 heats the hot plate 134 with an output amount according to the actual temperature of the hot plate 134.

[0041] The temperature sensor 139 is provided singly for the hot plate 134 and measures the temperature of the hot plate 134. The temperature sensor 139 may be provided inside the hot plate 134 or on the lower surface of the hot plate 134.

[0042] The support pins 135 are members that extend to penetrate the support table 131 and the heat plate 134 and support the wafer W from below. The support pins 135 move up and down to position the wafer W at a predetermined position. The support pins 135 are configured to transfer the wafer W between the temperature adjustment plate 151 that transports the wafer W and the support pins 135. For example, three support pins 135 are provided at equal intervals in the circumferential direction. The lifting mechanism 136 is configured to lift and lower the support pins 135 under the control of the controller 100.

[0043] The temperature adjustment mechanism 150 is configured to transfer (transfer) the wafer W between the hot plate 134 and an external transfer arm, and to adjust the temperature of the wafer W to a predetermined temperature. The temperature adjustment mechanism 150 includes a temperature adjustment plate 151 and a connecting bracket 152.

[0044] The temperature adjustment plate 151 is a plate that adjusts the temperature of the wafer W placed thereon, and more specifically, is a plate on which the wafer W heated by the heating plate 134 is placed and which cools the wafer W to a predetermined temperature. The temperature adjustment plate 151 is made of a metal with high thermal conductivity, such as aluminum, silver, or copper, and may be made of the same material from the viewpoint of preventing deformation due to heat, etc. A cooling flow path (not shown) is formed inside the temperature adjustment plate 151 to circulate cooling water and / or cooling gas.

[0045] The connecting bracket 152 is connected to the temperature adjustment plate 151 and is driven by a drive mechanism 153 controlled by the controller 100 to move within the housing 90. More specifically, the connecting bracket 152 is movable along a guide rail (not shown) that extends from the carry-in entrance 91 of the housing 90 to the vicinity of the heating mechanism 130. The movement of the connecting bracket 152 along the guide rail (not shown) allows the temperature adjustment plate 151 to move from the carry-in entrance 91 to the heating mechanism 130. The connecting bracket 152 is made of a metal with high thermal conductivity, such as aluminum, silver, or copper.

[0046] <Abnormality Detection Method> An abnormality detection method for the thermal processing unit U2 will now be described. This abnormality detection method detects abnormalities related to the hot plate 134 of the thermal processing unit U2, for example, detecting abnormalities in the temperature sensor 139 installed on the hot plate 134. If the accuracy of the temperature sensor 139 deteriorates over time, a discrepancy may occur between the true value and the temperature sensor indication, making it impossible to properly detect defects in the hot plate 134. For example, a method of detecting temperature sensor deterioration by duplicating the temperature sensor could be considered, but this has the drawback of being cumbersome and expensive. The abnormality detection method according to this embodiment was developed in consideration of the above-described circumstances, and detects deterioration of the temperature sensor 139 and properly detects defects related to the hot plate 134 without incurring additional complexity or cost.

[0047] The anomaly detection method includes a measurement step, a temperature integration step, a normal population creation step, and an anomaly detection step.

[0048] The measurement process is a process of measuring and storing the temperature change on the hot plate 134 during wafer processing. The controller 100 acquires and stores the temperature change on the hot plate 134 during wafer processing from the temperature sensor 139. FIG. 4(a) illustrates the temperature change on the hot plate 134 during the wafer heating process. In FIG. 4(a), the horizontal axis represents time, and the vertical axis represents the value measured by the temperature sensor 139 normalized by the set temperature (PV value). The integrated value of the temperature change during wafer processing is shown in the shaded area in FIG. 4(a). That is, the integrated value of the temperature change is represented by the temperature integrated area (shaded area) defined by the temperature curve showing the time-series curve of temperature change and the set temperature of the hot plate 134. In the measurement process, after placing the wafer W on the hot plate 134, the temperature change on the hot plate 134 is measured from at least the first time to the second time during which the temperature of the hot plate 134 changes.

[0049] The temperature integration process is a process of integrating the temperature changes during the processing of multiple wafers stored in the measurement process. The controller 100 integrates the stored temperature changes during the processing of multiple wafers. In the temperature integration process, as shown in FIG. 4( a), the temperature integration process calculates the temperature integration area defined by the range enclosed by the temperature curve showing the time-series curve of temperature change and the set temperature of the hot plate 134. Note that the temperature integration process may also integrate the temperature change by calculating a partial temperature integration area defined by the range enclosed by the set temperature of the hot plate 134 and the portion of the temperature curve from the peak of the temperature curve up to a predetermined time. That is, the temperature change may be integrated from a partial area rather than the entire area of ​​the shaded region shown in FIG. 4( a). FIG. 4( b) is a diagram illustrating the integration of the temperature changes of multiple wafers W at each set temperature. In FIG. 4( b), the horizontal axis represents the integrated value of temperature change (the temperature integrated area defined by the range (hatched area) enclosed by the temperature curve showing the time-series curve of temperature change and the set temperature of the heating plate 134). As shown in FIG. 4( b), the integrated values ​​of multiple wafers W at the same set temperature are generally similar. FIG. 4( c) is a diagram illustrating that the integrated value monotonically increases with the set temperature. In FIG. 4( c), the horizontal axis represents the set temperature, and the vertical axis represents the average integrated value. As shown in FIG. 4( c), the average integrated value monotonically increases with the set temperature.

[0050] In this embodiment, the difference from the start of monitoring is derived by monitoring the change in the characteristic value (the above-mentioned accumulated value) calculated from the PV value, and if the difference becomes large, an abnormality is detected as the temperature sensor 139 has deteriorated over time (details will be described later).

[0051] In the measurement process and the temperature integration process, the process may be changed depending on the type of wafer W. For example, when a warped wafer is heat-treated on the hot plate 134, the distance between the hot plate 134 and the wafer W is large, which may prevent the resist film from being heat-treated at the desired temperature, resulting in a reduced temperature drop and the inability to supply the desired amount of heat. The hatched area in FIG. 5( a) shows the temperature change during heat treatment of a normal wafer W. The hatched area in FIG. 5( b) shows the temperature change during heat treatment of a warped wafer. Because the amount of heat supplied to a warped wafer is clearly reduced, a feedback function may be provided depending on the type of wafer W, and the output and heat treatment time may be adjusted so that the amount of heat supplied to a warped wafer is equal to that of a normal wafer W. In the example shown in FIG. 5( c), the output and heat treatment time are adjusted for the warped wafer so that the amount of heat supplied to the warped wafer is equal to that of a normal wafer W. The wafer processing time on the heating plate 134 when the wafer W is a warped wafer may be set to be longer than the wafer processing time on the heating plate 134 when the wafer W is a normal wafer that is not a warped wafer.

[0052] Furthermore, in the measurement process and the temperature integration process, overshoot monitoring may be performed based on the PV value curve to determine whether or not an overshoot has occurred. Since overshoot monitoring is difficult to determine based on the integrated value alone, it may be performed by dividing the integrated value below the reference value into those above it and calculating their respective proportions in the total integrated value. Figure 6(a) shows a normal (no overshoot) PV value curve. In a normal PV value curve, the area above the reference value is nearly zero, and the upper integrated value / total integrated value is small (almost zero). Figure 6(b) shows a PV value curve in which an overshoot has occurred. In a PV value curve in which an overshoot has occurred, the upper integrated value / total integrated value is large (e.g., about 0.2). In such cases, it may be determined that an overshoot has occurred. If an overshoot has occurred, it may be due to an incorrectly set parameter or a wafer-related issue (such as a warped wafer).

[0053] In the normal population creation process, a normal population is created, which is a normal integration result for a predetermined number of wafers W integrated in the temperature integration process. The controller 100 may create a normal population, which is a normal integration result for a predetermined number of wafers W integrated. The normal population creation process may be executed when wafer processing on the hot plate 134 starts, or when wafer processing on the hot plate 134 is restarted after a change in the set temperature. Specifically, the normal population creation process may be executed when a new hot plate 134 is first used, after a change in the set temperature, immediately after an overhaul, or after maintenance. The normal population is created after confirming that the hot plate 134 being monitored is normal.

[0054] In the normal population creation step, a normal population may be created from normal integration results for a predetermined number of wafers W (e.g., 1,000 wafers) integrated in the temperature integration step, for example, at the start of wafer processing on the hot plate 134. In the normal population creation step, it may be determined whether the integration results are normally distributed, and if they are not, data may be reacquired. The determination of whether the distribution is normal may be performed, for example, by a Shapiro-Wilk test, and if the P value is less than the significance level, it may be determined that the distribution is not normal, and if the P value is equal to or greater than the significance level, it may be determined that the distribution is normal.

[0055] In the normal population creation process, the normal population may be created after deleting data on wafers W that fall outside a predetermined range based on the integrated value of temperature change. In this case, outlier data may be removed using, for example, a Smirnov-Grubbs (SG) test. Furthermore, in the normal population creation process, the created normal population may be compared with normal populations for other hot plates created for other hot plates, and an alert may be output if the difference between the created normal population and the normal populations for other hot plates is equal to or greater than a predetermined threshold, i.e., if the normal population is deemed inappropriate.

[0056] In the normal population creation process, sampling may be performed to include data for each of different product types of wafers W to create a normal population. For example, consider a case where product specifications change daily, as shown in FIG. 7( a). Assume that there are three types of wafers W, namely, product A, product B, and product C, and the operation is such that the product specifications change daily (a three-day cycle). In this case, as shown in FIG. 7( b), if a normal population is created using only wafers W for product A, for example, the integrated value for wafers W for product C may be an abnormal value (exceeding the threshold for detecting an abnormality, as described below). Therefore, as shown in FIG. 7( c), sampling is performed to include data (integrated values) for three types of wafers W, namely, product A, product B, and product C, to create a normal population. This reduces the risk of false detection for the three types of wafers W, as shown in FIG. 7( d).

[0057] In the normal population creation step, a normal population may be created by sampling wafers W of different product types. For example, consider a case where product specifications change monthly, as shown in FIG. 8( a). Assume that there are three types of wafers W, namely, product A, product B, and product C, and the product specifications change monthly (a three-month cycle). In cases where product specifications do not change over a relatively long period of one month, it is preferable to create a normal population for each product. That is, as shown in FIG. 8( b), a normal population is created by sampling to include only the data (integrated values) of wafers W of product A. As shown in FIG. 8( cb), a normal population is created by sampling to include only the data (integrated values) of wafers W of product B. As shown in FIG. 8( d), a normal population may also be created by sampling to include only the data (integrated values) of wafers W of product A. In this way, a normal population is created for each product specification, and anomaly detection is performed by comparing the data between products with the same specification (details will be described later), thereby improving the accuracy of anomaly detection. Alternatively, a determination may be made for a normal population for each product specification, and if the number of populations for each product specification is exceeded, an overall population (combining multiple product specifications) may be created.

[0058] The anomaly detection process creates an inspection object population, which is the integration result of the inspection objects of the predetermined number of wafers W integrated in the temperature integration process, and compares the inspection object population with a normal population to detect an anomaly related to the hot plate 134, specifically an anomaly of the temperature sensor 139. The controller 100 is configured to create an inspection object population, which is the integration result of the inspection objects of the predetermined number of wafers W integrated in the temperature integration process, and compares the inspection object population with a normal population to detect an anomaly related to the hot plate 134. The controller 100 detects an anomaly of the temperature sensor 139 as an anomaly related to the hot plate 134.

[0059] As described above, the normal population is a population created, for example, at the start of wafer processing on the hot plate 134 (when monitoring by the temperature sensor 139 begins). The test population is a population created, for example, some time after monitoring begins, and may be created, for example, at predetermined intervals. The normal population and the test population may be, for example, the cumulative results of wafers W processed on the hot plate 134 at the same set temperature and with a difference in number (e.g., the same number) within a predetermined range. Note that the normal population and the test population do not need to have the same number of wafers. If the normal population and the test population do not have the same number of wafers, an abnormality may be detected, for example, by performing a Welch t-test and then determining whether the P value is below the significance level. Regarding the difference in number, for example, the difference in the number of wafers that are outliers may be a fixed value, or there may be a threshold for that difference. For example, if the difference in number is, for example, 1 to 5% or more, the number of test substrates may be adjusted.

[0060] In the anomaly detection process, if the difference between the test population and the normal population exceeds a predetermined threshold, an anomaly related to the hot plate 134, specifically an anomaly in the temperature sensor 139, may be detected and an alert may be output. Whether the difference between the test population and the normal population exceeds a predetermined threshold may be determined, for example, by Welch's t-test. The Welch's t-test is a parametric test for the difference in mean values ​​between two data. As shown in FIG. 9(a), when there are two samples, the t-value shown by the formula in FIG. 9(b) is derived, and it is possible to determine that if the t-value is close to 0, there is no difference between the two samples, and if the t-value is far from 0, there is a difference between the two samples. In FIG. 9(b), X １ (Bar on top) is the average data of the test population, X ２ (Bar on top) is the mean value data of the normal population, N １ is the sample size of the test population, N ２ is the sample size of the normal population. １ is the unbiased standard deviation of the test population, S ２ is the unbiased standard deviation of a normal population. In this embodiment, a P value is calculated from the t value, and if the P value is less than the significance level, an abnormality in the temperature sensor 139 is detected.

[0061] As shown in FIG. 10 , for example, a normal population and a test population are prepared, each consisting of 1,000 wafers W processed on the hot plate 134 at the same set temperature (here, 80° C.). The data constituting the normal population and the test population are updated as appropriate. Then, the normal population and the test population, which are data on 1,000 wafers W, are compared, and a Welch's t-test, for example, is performed to determine a P value from the t value. If the P value is below the significance level, an alert is output. Note that the normal population and the test population do not need to be the same number of wafers.

[0062] Instead of setting a threshold value for the t value, an SPC chart may be used to detect an anomaly when the anomaly determination rule of the control chart is met, as shown in FIG. 11, for example.

[0063] In the anomaly detection process, if the difference between the test population and the normal population exceeds a predetermined threshold, a health check is performed to determine whether there has been a change in the test environment due to at least one of wafer-related and process-related factors. A process-related factor could be, for example, a process performed in a flow where the wafer is placed on the hot plate 134 and no bake process is performed. If the health check determines that there has been no change in the test environment due to wafer-related or process-related factors, an alert may be output.

[0064] The health check may involve anomaly detection (confirmation check) of a predetermined standard wafer (reference substrate). In this case, anomaly detection is performed by comparing a normal population of the prepared standard wafer with a population of standard wafers to be inspected. A Welch's t-test may also be performed. If the thickness of the wafer W is changed, the definition of the normal population may change from the start of monitoring due to wafer-related factors. The above method makes it possible to determine whether or not there has been a change in the inspection environment due to such wafer-related factors.

[0065] In the health check, anomaly detection may be performed without a substrate. Specifically, the top panel 132 is closed as shown in FIG. 12( a) and then opened as shown in FIG. 12( b) to temporarily lower the module temperature. Then, as shown in FIG. 12( c), the top panel 132 is closed again to return the temperature to the set value. Anomaly detection may be performed by comparing a pre-prepared normal population with the test population without a substrate. A Welch's t-test may also be performed. Alternatively, the heater 138 is switched from an on state to an off state and maintained in the off state to lower the module temperature. Then, the heater 138 is continuously turned on again, and anomaly detection may be performed by comparing a pre-prepared normal population with the test population without a substrate. A Welch's t-test may also be performed. Using such a method, it may be possible to determine whether or not there has been a change in the testing environment.

[0066] Next, the process of the abnormality detection method will be described with reference to FIGS.

[0067] 13 is a flowchart showing the overall flow of anomaly detection. First, a normal population is created (step S1). Specifically, for example, at the start of wafer processing on the hot plate 134 (when monitoring by the temperature sensor 139 begins), the temperature change during wafer processing is measured and stored (step S2), and an integrated value of the temperature change is calculated (step S3). More specifically, this is repeated until a predetermined number of wafers (e.g., 1,000 wafers) is reached (step S4). When the predetermined number of wafers is reached in step S4, the normal population creation process is completed, and the calculated integrated value is saved (step S5).

[0068] Next, anomaly detection is performed (step S6). Specifically, for example, after a certain period of time has elapsed since the start of monitoring, the temperature change during wafer processing is measured and stored (step S7), and the integrated value of the temperature change is calculated (step S8). More specifically, this is repeated until a predetermined number of wafers (e.g., 1,000 wafers) is reached (step S9).

[0069] When the predetermined number of sheets is reached in step S9, the integrated value of the normal population that has been saved in advance is input as data (step S10), the test population and the normal population are compared using Welch's t-test (step S11), and the P value is calculated from the t value. Then, it is determined whether the P value is below the significance level (step S12), and if it is not below the significance level, it is determined that there is no abnormality (step S13), and if it is below the significance level, it is determined that there is an abnormality (step S14).

[0070] 14 is a flowchart showing the details of creating a normal population. When the creation of a normal population is started (step S1), temperature measurement recording (step S2) and integrated value calculation (step S3) are repeated until a predetermined number of sheets (e.g., 1,000 sheets) is reached (step S4). Then, for example, a Shapiro-Wilk test (a test to confirm normal distribution) is performed on the integrated results (step S101), and it is determined whether the P value is equal to or greater than the significance level (step S102). If the P value is not equal to or greater than the significance level in step S102, data is re-collected. If the P value is equal to or greater than the significance level, it is determined to be a normal distribution (step S103).

[0071] Next, an outlier test is performed to remove data for wafers W that fall outside a predetermined range of integrated temperature changes (step S104). For example, data that falls outside the mean ±3σ is removed (step S105). Alternatively, outliers may be removed using an SG test. Furthermore, the created normal population is compared with other normal populations created for other hot plates (oven modules) (step S106), and it is determined whether the difference exceeds a predetermined threshold (step S107). If the difference exceeds the threshold, data is re-collected. If the difference does not exceed the threshold, it is determined that there is no problem with the normal population (step S108). The normal population is then saved, and the process ends (step S109).

[0072] 15 is a flowchart showing the details of anomaly detection. When anomaly detection is initiated (step S6), temperature measurement recording (step S7) and integrated value calculation (step S8) are repeated until a predetermined number of sheets (e.g., 1,000 sheets) is reached (step S9). Then, a normal population is input from the stored data (step S10), and the test population and the normal population are compared using Welch's t-test (step S11), and a P value is calculated from the t value. Then, it is determined whether the P value is below the significance level (step S12). If not, it is determined that there is no anomaly (step S13). If it is below the significance level, a health check is initiated (step S201).

[0073] When a reference wafer is used as a health check, if the P value is not below the significance level, for example, it is fed back to the control (step S203).On the other hand, if it is below the significance level (if there is no change in the inspection environment due to the wafer), an alert is output (step S204).

[0074] Next, the effects of the anomaly detection method according to this embodiment will be described.

[0075] The anomaly detection method according to this embodiment is an anomaly detection method relating to the hot plate 134 for processing wafers W, and includes a measurement step of measuring and storing temperature changes during wafer processing on the hot plate 134. It also includes a temperature integration step of integrating the temperature changes during processing of a plurality of wafers stored in the measurement step, and a normal population creation step of creating a normal population, which is a normal integration result for a predetermined number of wafers W integrated in the temperature integration step. It also includes an anomaly detection step of creating an inspection target population, which is an integration result for inspection targets for a predetermined number of wafers W integrated in the temperature integration step, and comparing the inspection target population with the normal population to detect an anomaly relating to the hot plate 134.

[0076] This anomaly detection method can detect anomalies based on the difference between the time when the normal population was created (when monitoring began) and the current state. In wafer heat treatment processes, the set temperature has a significant impact (such as the temperature drop during wafer loading and the time it takes to recover to the set temperature). By using integrated values, it is possible to more accurately capture state differences than when using a single data set. For example, it is possible to easily monitor temperature sensor degradation over time, which was previously impossible to detect or required duplication. If the accuracy of a temperature sensor deteriorates over time, a discrepancy may occur between the true value and the temperature sensor reading, potentially making it impossible to properly detect a fault in the hot plate. While a method for detecting temperature sensor degradation, such as by using duplicate temperature sensors, is conceivable, this method is complex and expensive. In this regard, this anomaly detection method can properly detect temperature sensor degradation and, ultimately, faults in the hot plate 134, without the need for duplicate temperature sensors. The present invention can also be applied to detecting abnormalities in the heater 138 itself, abnormalities related to its control, foreign matter caught between the wafer W and the heating plate 134, and the like.

[0077] In the anomaly detection process, if the difference between the test population and the normal population exceeds a predetermined threshold, an anomaly related to the heating plate 134 may be detected and an alert may be output. This allows appropriate setting of the specific level of difference required to detect an anomaly and output an alert.

[0078] The normal population and the test population may be the cumulative results of wafers W with a difference in number within a predetermined range (e.g., the same number) that have been processed at the same set temperature on the heating plate 134. By comparing wafers W with a difference in number within a predetermined range and at the same set temperature in this manner, deterioration of the temperature sensor 139 can be more appropriately detected.

[0079] The normal population creation process may be performed when wafer processing on the heating plate 134 starts, or when wafer processing on the heating plate 134 is restarted after a change in the set temperature. This allows the normal population to be created appropriately when the temperature sensor 139 is not yet deteriorated.

[0080] In the normal population creation step, the normal population may be created after deleting data of wafers W that are outside a predetermined range from the integrated value of the temperature change obtained in the measurement step. This allows the normal population to be created appropriately by excluding outliers.

[0081] In the normal population creation process, the created normal population may be compared with other hot plate normal populations created for other hot plates, and an alert may be output if the difference between the created normal population and the other hot plate normal population is equal to or greater than a predetermined threshold. In this way, if a normal population that is clearly inappropriate compared with data from other hot plates is created, for example, the normal population can be recreated.

[0082] In the normal population creation step, it may be determined whether the integration results are normally distributed, and if they are not, the data may be reacquired, thereby allowing an appropriate normal population to be created.

[0083] In the anomaly detection process, when the difference between the test population and the normal population exceeds a predetermined threshold, a health check is performed to determine whether or not there has been a change in the test environment due to at least one of the wafer W and the processing. If the health check determines that there has been no change in the test environment due to the wafer W, an alert may be output. This makes it possible to appropriately output an alert when there has been no change in the test environment due to the wafer, that is, when an abnormality in the temperature sensor 139 or the like is suspected.

[0084] In the normal population creation step, the normal population may be created so as to include data on wafers W of different product types. For example, in cases where product specifications change over a short period of time, if the normal population is created based only on wafers of a single product type, there is a risk that data on wafers of other types will be determined to be abnormal. In this regard, by creating the normal population so as to include data on wafers W of different product types, the normal population can be created taking into account the data on wafers W of each product type. This makes it possible to prevent a determination that a wafer W is abnormal simply because it is a different product type.

[0085] In the normal population creation step, a normal population may be created for each different product type of wafer W. For example, when product specifications change only over a relatively long time span, a normal population is created for each product type, and the normal population and the test population are compared for the same product type, thereby enabling more accurate detection of abnormalities.

[0086] The wafer processing time on the heating plate 134 when the wafer W is a warped wafer may be set longer than the wafer processing time on the heating plate 134 when the wafer W is a normal substrate that is not a warped wafer. This allows the temperature change to be the same even for a warped wafer to which the desired amount of heat cannot be supplied, thereby improving process uniformity.

[0087] A single temperature sensor 139 may be provided for the hot plate 134. This configuration reduces the complexity of processing and reduces costs compared to a configuration in which duplicated temperature sensors are provided. Furthermore, even if a single temperature sensor 139 is provided, the above-described processing can be used to detect deterioration of the temperature sensor 139 itself without any problems.

[0088] Furthermore, in the measurement step, after placing a substrate on the hot plate 134, the temperature change of the hot plate 134 may be measured at least from a first time to a second time during which the temperature of the hot plate 134 changes. In the temperature integration step, an integrated temperature area may be calculated, which is defined by the range enclosed by a temperature curve showing a time-series curve of temperature change and the set temperature of the hot plate 134. The temperature change may then be integrated based on the integrated temperature area. This allows for highly accurate calculation of the integrated temperature change value. In addition, in the temperature integration step, the temperature change may be integrated by deriving a partial integrated temperature area, which is defined by the range enclosed by the set temperature of the hot plate 134 and the portion of the temperature curve from the peak of the temperature curve to a predetermined time. This allows for appropriate calculation of the integrated temperature change value while avoiding data congestion. In addition, in the health check of the anomaly detection step, anomaly detection may be performed on a specified standard wafer or without a wafer W. This allows for appropriate determination of whether an anomaly is caused by a wafer or the like. The anomaly detection method according to the present embodiment has been described above, but the anomaly detection method of the present disclosure is not limited to the above. For example, although an anomaly determination has been described as being based on the integrated value of the PV value, an anomaly determination may also be performed based on, for example, the minimum value of the PV value. Furthermore, in order to reduce the data load, an anomaly determination may also be performed based on only a portion of the data of the integrated value of the PV value (specifically, for example, up to the position after the peak). Furthermore, an anomaly determination may also be performed based on the MV value instead of the PV value.

[0089] The present disclosure is not limited to the above-described embodiment. For example, the minimum PV value obtained during the heat treatment of the wafer W may be calculated, the transition of the most recent mode (of the test population) may be monitored, and the most recent mode (of the test population) may be compared (e.g., the difference may be calculated) with the mode at the start of monitoring (of the normal population). An abnormality related to the heating plate may then be detected based on the comparison result. In this way, in the abnormality detection process, an abnormality related to the heating plate may be detected by comparing the mode in the test population with the mode in the normal population (first modification).

[0090] Specifically, anomaly detection may be performed based on statistics of the data set at the start of monitoring (normal population), predetermined thresholds, and anomaly detection rules for the control chart. In this case, anomaly or normality is determined based on whether the anomaly detection rule is met. If an anomaly is determined based on the anomaly detection rule, it can be determined that the corresponding temperature sensor is likely to have deteriorated, causing a deviation from the true value. At customer sites, wafers W in various states are expected to be loaded into the hot plate module, and the distribution of features derived from the PV values ​​is expected to differ for each. The anomaly detection method using the t-test described above assumes that when wafers W are loaded onto the hot plate in a single or multiple states, the proportion of these states is fixed. When this assumption does not hold, false detection occurs. This method assumes that wafers W in various states are loaded into the hot plate module and employs appropriate statistics to achieve highly robust anomaly detection.

[0091] FIG. 16 is a graph illustrating the anomaly detection method according to the first modified example, showing an SPC chart. FIG. 16 shows a graph with the mean value on the vertical axis, a graph with the median value on the vertical axis, and a graph with the mode value on the vertical axis, all of which show time on the horizontal axis. In the SPC chart of FIG. 16 , the range of the upper and lower horizontal lines (dashed lines) is determined to be normal, while the range outside the range of the upper and lower horizontal lines (dashed lines) is determined to be abnormal. For example, at the timing indicated by the vertical lines (dashed lines) on the left side of FIG. 16 , the mean and median of the minimum PV values ​​are large and fall within the abnormal range, while the mode is near the center and falls within the normal range. Thus, the mean and median are susceptible to outliers, while the mode is less susceptible to outliers. By performing an anomaly detection based on such a mode, the influence of outliers can be suppressed, enabling highly accurate anomaly detection. Note that at the timing indicated by the vertical lines (dashed lines) on the right side of FIG. 16 , the mean, median, and mode of the minimum PV values ​​are all within the abnormal range. By using the SPC chart to determine whether an abnormality exists at a number of consecutive points, rather than at a single point, the accuracy of the determination can be increased. By performing the determination based on the most frequent value, the risk of erroneous determination can be reduced.

[0092] FIG. 17 is a graph illustrating an anomaly detection method according to a second modification, showing an SPC chart. Here, wafers W loaded onto a certain heating plate may be transferred using various transfer recipes. Because the state of the wafer W is generally linked to the transfer recipe, differences in the transfer recipe may result in differences in the distribution of feature quantities calculated from the PV values. As shown in the top graph of FIG. 17 , if average values ​​are plotted without distinguishing between two transfer recipes ("Transfer Recipe 1" and "Transfer Recipe 2"), wafers W in various states may be loaded. As a result, an SPC chart may determine an anomaly even when there is no actual anomaly. In this regard, as shown in the vertically middle and bottom graphs of FIG. 17 , the average values ​​are plotted separately for the two transfer recipes ("Transfer Recipe 1" and "Transfer Recipe 2"). In this case, false detections that would occur when plotting multiple transfer recipes without distinguishing between them, as described above, can be suppressed. To realize such an anomaly detection method, a normal population may be created for each transfer recipe in the normal population creation step, and an anomaly detection step may detect an anomaly related to the heating plate by comparing the test population with a normal population corresponding to the transfer recipe of the test population. By grouping wafers with the same transfer recipe, the wafer state becomes uniform, which is expected to reduce variation.

[0093] FIG. 18 is a graph illustrating an anomaly detection method according to a third modified example, showing an SPC chart. The upper graph in FIG. 18 is an SPC chart for a certain hot plate module ("Module a"), with the vertical axis representing, for example, the average value and the horizontal axis representing time. The lower graph in FIG. 18 is an SPC chart for a certain hot plate module ("Module b") other than "Module a", with the vertical axis representing, for example, the average value and the horizontal axis representing time. In the normal population creation process, a normal population is created for each transfer recipe for each module including a hot plate (here, for each of "Module a" and "Module b"). Both the upper and lower graphs in the SPC chart in FIG. 18 show the results of a certain identical transfer recipe ("recipe 1"). In such a case, the anomaly detection process may determine whether the difference between the test population and the normal population for the same transfer recipe in different modules exceeds a predetermined threshold. As shown in FIG. 18 , if the PV values ​​for the same transfer recipe ("recipe 1") in each of two modules become abnormal, it is inferred that the wafer state for that transfer recipe has changed. In this case, a notification may be output to prompt the re-creation of a normal population (model reconstruction). In this way, if abnormalities are detected for the same transfer recipe in multiple modules at the same time, it is inferred that the state of the transferred wafers W has changed (it is believed that the feature distribution has changed due to the recipe). Therefore, by prompting the re-construction of the reference, false detections can be suppressed. For example, for a given transfer recipe ("recipe 1"), as shown in FIG. 19 , if the PV value becomes abnormal only in one module ("Module a"), deterioration of the temperature sensor may be suspected, and an abnormal alarm may be output for that module ("Module a"). Furthermore, if the PV value becomes abnormal only in the other module ("Module b"), deterioration of the temperature sensor may be suspected, and an abnormal alarm may be output for the other module ("Module b").

[0094] 20 is a diagram illustrating an anomaly detection method according to a fourth modified example. In the normal population creation process, a normal population is created for each transfer recipe for each module including a hot plate. In the anomaly detection process, an anomaly related to the hot plate may be detected when the difference between the test population and the normal population for different transfer recipes in the same module exceeds a predetermined threshold. In this way, an anomaly related to the hot plate is detected only when an anomaly is detected in multiple transfer recipes in the same module, rather than just one. For example, an anomaly alarm is output, enabling more accurate anomaly detection using multiple transfer recipes.

[0095] Finally, various exemplary embodiments included in the present disclosure are described below in [E1] to [E15].

[0096] [E1] An anomaly detection method for a hot plate for processing substrates, comprising: a measurement step of measuring and storing temperature changes during substrate processing on the hot plate; a temperature accumulation step of accumulating the temperature changes during processing of a plurality of substrates stored in the measurement step; a normal population creation step of creating a normal population which is a normal accumulation result for a predetermined number of substrates accumulated by the temperature accumulation step; and an anomaly detection step of creating a test subject population which is an accumulation result for test subjects for a predetermined number of substrates accumulated by the temperature accumulation step, and comparing the test subject population with the normal population to detect an anomaly related to the hot plate.

[0097] [E2] The anomaly detection method according to [E1], wherein, in the anomaly detection step, when a difference between the test population and the normal population exceeds a predetermined threshold, an anomaly related to the hot plate is detected and an alert is output.

[0098] [E3] An anomaly detection method according to [E1] or [E2], wherein the normal population and the test population are the cumulative results of substrates with a difference in number within a predetermined range that have been subjected to substrate processing on the hot plate at the same set temperature.

[0099] [E4] The abnormality detection method according to any one of [E1] to [E3], wherein the normal population creation process is executed at the start of substrate processing on the hot plate or at the restart of substrate processing on the hot plate after changing the set temperature.

[0100] [E5] An anomaly detection method described in any one of [E1] to [E4], in which in the normal population creation process, a normal population is created after deleting data on substrates that fall outside a predetermined range from the integrated value of the temperature change obtained in the measurement process.

[0101] [E6] In the normal population creation process, the normal population that is created is compared with the normal population of other hot plates that are created for other hot plates, and when the difference between the normal population of other hot plates and the normal population of other hot plates is greater than or equal to a predetermined threshold, an alert is output.

[0102] [E7] An anomaly detection method according to any one of [E1] to [E6], wherein in the normal population creation process, it is determined whether the accumulation result is normally distributed, and if it is not normally distributed, the data is reacquired.

[0103] [E8] In the anomaly detection step, when the difference between the test population and the normal population exceeds a predetermined threshold, a health check is performed to determine whether or not there has been a change in the test environment due to at least one of the substrate and the processing. If it is determined in the health check that there has been no change in the test environment due to the substrate, an alert is output.

[0104] [E9] The anomaly detection method according to any one of [E1] to [E8], wherein in the normal population creation step, the normal population is created so as to include data for each of different product types of boards.

[0105] [E10] The anomaly detection method according to any one of [E1] to [E8], wherein in the normal population creation step, the normal population is created for each of different product types of boards.

[0106] [E11] The abnormality detection method according to any one of [E1] to [E10], wherein the substrate processing time on the hot plate when the substrate is a warped wafer is set longer than the substrate processing time on the hot plate when the substrate is a normal substrate that is not a warped wafer.

[0107] [E12] The anomaly detection method according to any one of [E1] to [E11], wherein the anomaly detection step detects an anomaly in a temperature sensor installed on the hot plate.

[0108] [E13] The anomaly detection method according to [E12], wherein the temperature sensor is provided singly for the hot plate.

[0109] [E14] In the measuring step, after placing the substrate on the hot plate, measure the temperature change of the hot plate from the first time to the second time when the temperature of the hot plate changes; In the temperature integrating step, derive the temperature integrating area that is determined by the range surrounded by the temperature curve that shows the time series curve of the temperature change and the set temperature of the hot plate, thereby integrating the temperature change.

[0110] [E15] In the temperature integrating process, the temperature integrating area is determined by the range that is defined by the part of the temperature curve that is past the peak of the temperature curve and the predetermined time and the set temperature of the hot plate, and thus the abnormality detecting method described in [E14] is carried out to integrate the temperature change.

[0111] [E16] The anomaly detection method according to [E8], wherein the health check in the anomaly detection step detects anomalies on a predetermined reference substrate or in a state where no substrate is present.

[0112] [E17] A substrate processing apparatus comprising: a hot plate on which a substrate is placed and which applies heat to the substrate; a temperature regulator which heats the hot plate; a temperature sensor which measures the temperature of the hot plate; and a control unit, wherein the control unit is configured to: obtain from the temperature sensor and store temperature changes on the hot plate during substrate processing; add up the stored temperature changes during processing of a plurality of substrates; create a normal population which is a normal addition result for the added up predetermined number of substrates; and create an inspection target population which is an addition result for inspection targets for the added up predetermined number of substrates, and detect an abnormality related to the hot plate by comparing the inspection target population with the normal population.

[0113] [E18] The substrate processing apparatus according to [E14], wherein the control unit detects an abnormality in the temperature sensor as an abnormality related to the hot plate.

[0114] [E19] The anomaly detection method according to any one of [E1] to [E18], wherein the anomaly detection step detects an anomaly related to the hot plate by comparing the most frequent value in the test population with the most frequent value in the normal population.

[0115] [E20] An anomaly detection method according to any one of [E1] to [E19], wherein in the normal population creation step, the normal population is created for each transfer recipe, and in the anomaly detection step, an anomaly related to a hot plate is detected by comparing the test population with the normal population corresponding to the transfer recipe of the test population.

[0116] [E21] The anomaly detection method described in [E20], wherein the normal population creation process creates the normal population for each transfer recipe for each module including a hot plate, and the anomaly detection process outputs a notification urging the user to recreate the normal population when a difference between the test population and the normal population for the same transfer recipe in different modules exceeds a predetermined threshold.

[0117] [E22] The anomaly detection method described in [E20], wherein the normal population creation step creates the normal population for each transfer recipe for each module including a hot plate, and the anomaly detection step detects an anomaly related to the hot plate when the difference between the test population and the normal population for different transfer recipes in the same module exceeds a predetermined threshold.

[0118] 1...wafer processing system (substrate processing apparatus), 134...hot plate, 100...controller (control unit), 138...heater (temperature regulator), 139...temperature sensor, W...wafer (substrate).

Claims

1. A method for detecting an abnormality related to a hot plate used to process substrates, comprising: a measurement step of measuring and storing temperature changes during substrate processing on the hot plate; a temperature accumulation step of accumulating the temperature changes during processing of multiple substrates stored in the measurement step; a normal population creation step of creating a normal population that is a normal accumulation result for a predetermined number of substrates accumulated in the temperature accumulation step; and an abnormality detection step of creating a test subject population that is an accumulation result for test subjects of a predetermined number of substrates accumulated in the temperature accumulation step, and comparing the test subject population with the normal population to detect an abnormality related to the hot plate.

2. The anomaly detection method according to claim 1, wherein the anomaly detection step detects an anomaly related to the hot plate and outputs an alert when the difference between the test population and the normal population exceeds a predetermined threshold.

3. The anomaly detection method according to claim 1, wherein the normal population and the test population are the cumulative results of substrates processed on the heating plate at the same set temperature, with the difference in number of substrates being within a predetermined range.

4. The anomaly detection method according to claim 1, wherein the normal population creation step is performed when substrate processing on the heating plate starts or when substrate processing on the heating plate is resumed after a set temperature has been changed.

5. An anomaly detection method as described in claim 1, wherein in the normal population creation step, a normal population is created after deleting data of substrates that fall outside a predetermined range from the integrated value of temperature change obtained in the measurement step.

6. The anomaly detection method according to claim 1, wherein in the normal population creation step, the created normal population is compared with other hot plate normal populations created for other hot plates, and an alert is output if the difference between the created normal population and the other hot plate normal populations is equal to or greater than a predetermined threshold.

7. The anomaly detection method according to claim 1, wherein in the normal population creation step, it is determined whether the accumulation result is normally distributed, and if it is not normally distributed, data is reacquired.

8. The anomaly detection method according to claim 2, wherein the anomaly detection step outputs an alert when the difference between the test population and the normal population exceeds a predetermined threshold, and when a health check is performed to determine whether or not there has been a change in the test environment due to at least one of the substrate and the processing, and when it is determined that there has been no change in the test environment due to the substrate.

9. The anomaly detection method according to claim 1, wherein in the normal population creation step, the normal population is created so as to include data on boards of different product types.

10. The anomaly detection method according to claim 1, wherein in the normal population creation step, the normal population is created for each different product type of board.

11. The anomaly detection method according to claim 1, wherein the substrate processing time on the heating plate when the substrate is a warped wafer is set longer than the substrate processing time on the heating plate when the substrate is a normal substrate that is not a warped wafer.

12. The anomaly detection method according to any one of claims 1 to 11, wherein the anomaly detection step detects an anomaly in a temperature sensor installed on the hot plate.

13. The anomaly detection method according to claim 12, wherein the temperature sensor is provided singly for the hot plate.

14. An anomaly detection method as described in claim 1, wherein in the measuring step, after placing a substrate on the hot plate, the temperature change of the hot plate is measured at least from a first time to a second time when the temperature of the hot plate changes, and in the temperature integrating step, the temperature change is integrated by deriving a temperature integration area determined by the range surrounded by a temperature curve showing a time-series curve of the temperature change and the set temperature of the hot plate.

15. An anomaly detection method as described in claim 14, wherein the temperature integration step integrates the temperature change by deriving a partial temperature integration area determined by the range enclosed by the portion of the temperature curve from the peak of the temperature curve up to a predetermined time and the set temperature of the heating plate.

16. The anomaly detection method according to claim 8, wherein the health check in the anomaly detection step detects anomalies in a predetermined reference substrate or in a state where no substrate is present.

17. A substrate processing apparatus comprising: a hot plate on which a substrate is placed and which applies heat to the substrate; a temperature regulator which heats the hot plate; a temperature sensor which measures the temperature of the hot plate; and a control unit, wherein the control unit is configured to: acquire from the temperature sensor and store temperature changes on the hot plate during substrate processing; add up the stored temperature changes during processing of a plurality of substrates; create a normal population which is the normal addition result for the added up predetermined number of substrates; and create an inspection target population which is the addition result for inspection targets for the added up predetermined number of substrates, and detect an abnormality related to the hot plate by comparing the inspection target population with the normal population.

18. The substrate processing apparatus according to claim 17, wherein the control unit detects an abnormality in the temperature sensor as an abnormality related to the heating plate.

19. An anomaly detection method according to claim 1, wherein the anomaly detection step detects an anomaly related to the hot plate by comparing the most frequent value in the test population with the most frequent value in the normal population.

20. An anomaly detection method as described in claim 1, wherein in the normal population creation step, the normal population is created for each transport recipe, and in the anomaly detection step, an anomaly related to the hot plate is detected by comparing the inspection target population with the normal population corresponding to the transport recipe of the inspection target population.

21. An anomaly detection method as described in claim 20, wherein the normal population creation step creates the normal population for each transfer recipe for each module including a hot plate, and the anomaly detection step outputs a notification urging the user to recreate the normal population when the difference between the test population and the normal population for the same transfer recipe in different modules exceeds a predetermined threshold.

22. An anomaly detection method as described in claim 20, wherein in the normal population creation step, the normal population is created for each transfer recipe for each module including a hot plate, and in the anomaly detection step, an anomaly related to the hot plate is detected when the difference between the inspection target population and the normal population for different transfer recipes in the same module exceeds a predetermined threshold.

Images