Liquid treatment method, liquid treatment apparatus, and program

Abstract

A liquid processing method of the present disclosure performs: a step for circulating a processing liquid in a flow path connecting a storage part in which the processing liquid is stored and a processing liquid supply part to discharge the processing liquid from the processing liquid supply part to a substrate, and processing the substrate; a retention step for filling and retaining liquid in the flow path; a circulation step for circulating the retained liquid in the flow path; a signal acquisition step for performing light irradiation on the flow path with a light irradiation part during the circulation step, and light reception of light from the flow path with a light reception part, and acquiring a detection signal corresponding to foreign matter in the liquid output from the light reception part; and a responding step for, on the basis of the detection signal, estimating an abnormal location in the flow path, presenting a countermeasure against the abnormality, or executing the countermeasure.

Description

Technical Field

[0001] The present disclosure relates to a liquid processing method, a liquid processing apparatus, and a program.

Background Art

[0002] In the manufacturing process of semiconductor devices, a processing liquid is supplied to a semiconductor wafer (hereinafter referred to as a wafer) to perform processes such as film formation and cleaning. Patent Document 1 discloses a technique for optically detecting foreign substances in the processing liquid supplied to a wafer.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] The present disclosure provides a technique capable of preventing an increase in the time during which the apparatus cannot be operated due to foreign substances in the flow path of the processing liquid when operating a liquid processing apparatus that supplies a processing liquid to a substrate.

Means for Solving the Problems

[0005] The liquid processing method of the present disclosure includes a step of flowing the processing liquid through a flow path connecting a storage unit in which the processing liquid is stored and a processing liquid supply unit, discharging the processing liquid from the processing liquid supply unit to a substrate, and processing the substrate; a retention step of filling the flow path with a liquid and retaining the liquid; a flowing step of flowing the retained liquid through the flow path; a signal acquisition step of irradiating the flow path with light by a light irradiation unit and receiving light from the flow path by a light receiving unit while performing the flowing step, and acquiring a detection signal corresponding to foreign substances in the liquid output from the light receiving unit; Based on the detection signal, a corresponding process of estimating an abnormal location in the flow path, presenting a countermeasure operation for the abnormality, or implementing the countermeasure operation is performed. It includes.

Effect of the Invention

[0006] When operating a liquid processing apparatus that supplies a processing liquid to a substrate, the present disclosure can prevent the time during which the apparatus cannot be operated from becoming long due to foreign matter in the flow path of the processing liquid.

Brief Description of the Drawings

[0007]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Embodiments for Carrying Out the Invention

[0008] A resist coating apparatus 1 which is an embodiment of the liquid processing apparatus of the present disclosure will be described with reference to FIG. 1. The resist coating apparatus 1 spins and coats a resist, which is a processing liquid, on a wafer W, which is a substrate, to form a resist film. The resist coating apparatus 1 includes a processing unit 11, a piping system 2, a foreign matter detection mechanism 3, and a control unit 4.

[0009] The above-mentioned processing unit 11 includes a spin chuck 12, a rotation mechanism 13, and a cup 14. A transfer mechanism (not shown) outside the resist coating apparatus 1 delivers the wafer W to the spin chuck 12. The back central portion of the wafer W is sucked and held by the spin chuck 12, and the spin chuck 12 is rotated by the rotation mechanism 13. The cup 14 surrounds the side circumference of the wafer W held by the spin chuck 12 and receives the resist scattered from the wafer W.

[0010] Subsequently, the configuration of the piping system 2 will be described. In order to avoid complication of the description, the piping system 2 is shown in FIG. 1 in a simplified manner compared to the actual apparatus configuration. The piping system 2 includes a nozzle 21. The nozzle 21 is movable between a processing position above the central portion of the wafer W held by the spin chuck 12 and a standby position outside the cup 14 by a moving mechanism (not shown), and discharges the resist downward.

[0011] When performing processing on the wafer W, the nozzle 21 moves from the standby position to the processing position and discharges the resist onto the central portion of the wafer W. The wafer W onto which the resist has been discharged by the rotation mechanism 13 rotates, and the resist, which is a processing liquid, is applied to the entire surface of the wafer W, and a resist film is formed from the resist. That is, the processing step of the substrate is performed. Note that when the wafer W is not being processed, thinner may be discharged from the nozzle 21 as a liquid other than the resist, which will be described in detail later.

[0012] In addition to the above-mentioned nozzle 21, the piping system 2 includes a pipe 22, valves V1 to V4, a pump 23, a filter 24, a resist storage section 25, a pressurizing pipe 26, a supply source 27 of N2 (nitrogen) gas, and a discharge pipe 18. The downstream end of the pipe 22 is connected to the nozzle 21. The upstream side of the pipe 22 is connected to a resist storage section 25 in which the resist is stored via the valve V1, the pump 23, the valve V2, and the filter 24 in this order. The pipe 22 and each member interposed in the pipe 22 form a flow path connecting the resist storage section 25, which is a storage section for the processing liquid, and the nozzle 21, which is a processing liquid supply section.

[0013] When the valve V1 is closed and the valve V2 is opened, the resist in the resist storage unit 25 flows downstream through the pipe 22 by the suction operation of the pump 23 and is stored in the pump 23. Then, when the valve V1 is opened and the valve V2 is closed, the resist is supplied from the pump 23, which is a flow-through mechanism, to the downstream side of the pipe system 2 by the discharge operation of the pump 23 (the operation of squeezing out the liquid from the pump 23) and is discharged from the nozzle 21.

[0014] The discharge amount in each discharge operation of the pump 23 is set to a predetermined value. The control unit 4 described later is a computer, and based on the set value of the discharge amount in this one discharge operation and the number of discharge operations, it is configured to be able to calculate the discharge amount of the liquid (resist or thinner) discharged from the nozzle 21 by the operation of the pump 23 from an arbitrary point in time. Therefore, the pump 23 and the control unit 4 constitute a discharge amount measuring mechanism.

[0015] Regarding the flow path through which the resist flows from the tip of the nozzle 21 to the upstream end of the pipe 22, it is defined as the flow path 20. Further, the downstream side of the pressurizing pipe 26 is connected to the resist storage unit 25, and the pressurizing pipe 26 opens into the gas phase in the resist storage unit 25. The upstream end of the pressurizing pipe 26 is connected to the N2 gas supply source 27 via the valve V3. The supply and cut-off of N2 gas to the resist storage unit 25 are controlled by opening and closing the valve V3. By supplying the N2 gas, the inside of the resist storage unit 25 is pressurized and the resist is pressure-fed to the downstream side of the pipe 22. Note that the pressure feeding of the resist by this N2 gas is performed when performing the cleaning operation of the flow path 20 described later, and when discharging the resist onto the wafer W described above, the discharge operation of the pump 23 is performed.

[0016] The resist storage unit 25 can be replaced with a thinner storage unit 28 in which thinner is stored instead of the resist. When starting up the apparatus described later, in order to clean the piping system 2, the piping 22 and the pressurizing piping 26 are connected to the thinner storage unit 28 in this way. At that time, instead of the resist, the thinner is supplied from the thinner storage unit 28 to the piping 22 and discharged from the nozzle 21. Also, when starting up the apparatus, the above-described filter 24 is removed from the piping system 2 in order to increase the fluidity of the liquid and improve the cleanability. At that time, by attaching a flow path forming member 19 having no filter function instead of the filter 24, the liquid can flow through the piping system 2.

[0017] One end of the discharge piping 18 is connected to the pump 23, and the other end of the discharge piping 18 opens to the atmospheric atmosphere. A valve V4 is provided in the discharge piping 18. After the pump 23 performs the liquid suction operation as described above, the valves V1 and V2 are closed, and the pump 23 operates with the valve V4 opened, so that the bubbles accumulated in the pump 23 are removed through the discharge piping 18. The valve V4 is closed except when discharging these bubbles.

[0018] A light transmission part 31 is provided near the nozzle 21 on the downstream side of the valve V1 of the piping 22. More specifically, a part of the flow path from the valve V1 to the nozzle 21 is formed by the light transmission part 31, and a part of the above-described flow path 20 is constituted by the light transmission part 31. The light transmission part 31 is made of, for example, quartz so that the foreign matter detection mechanism 3 can optically detect foreign matters.

[0019] Hereinafter, the foreign matter detection mechanism 3 will be described. The foreign matter detection mechanism 3 includes a light projecting part 32 and a light receiving part 33, and the light projecting part 32 and the light receiving part 33 are respectively provided outside the light transmission part 31. The light projecting part 32 irradiates light toward the flow path formed by the light transmission part 31. The light receiving part 33 is arranged so as to be able to receive the lateral scattered light generated in the flow path of the light transmission part 31, and outputs a detection signal corresponding to the received light to the control part 4.

[0020] For each wafer W, one discharge operation of the pump 23 is performed and the above-described coating process is carried out. During this one discharge operation, that is, during the period when the resist flows through the flow path of the light transmission portion 31, light irradiation from the light projection portion 32 to the light transmission portion 31 and acquisition of a detection signal by the control unit 4 are performed. Therefore, data (i.e., time-series data) regarding the transition (behavior) of the signal level during that period is acquired. And this time-series data includes information regarding the number and particle size of particles that are foreign substances in the liquid. That is, the foreign substance detection mechanism 3 is configured to output a detection signal corresponding to the foreign substances in the liquid. The control unit 4 determines the propriety of the process on the wafer W based on the above time-series data.

[0021] As will be described in detail later, the foreign substance detection mechanism 3 also detects particles other than during the processing of the wafer W. Also in that case, the detection signal is acquired for a predetermined period so as to form time-series data. In addition to the particles in the resist, the foreign substance detection mechanism 3 is also used to detect the particles in the thinner. According to the foreign substance detection mechanism 3, it is possible to detect particles having a particle size of, for example, 10 nm to 200 nm. In each step for detecting the number of particles in each of the work flows described later, the number of particles having a particle size within a predetermined range within the detectable range is detected. Note that the particles detected here include, for example, bubbles and those having an abnormal size among the components of the resist such as polymers.

[0022] By the way, an abnormality has occurred at a specific location in the resist flow path in the piping system 2, and in some cases, a relatively large number of the above-mentioned particles may be generated from the location (for convenience, referred to as the contaminated location). That is, some of the members constituting the flow path such as the valve V1 and the pump 23 described above may become the contaminated location. By performing a cleaning operation on the flow path, the flow path can be cleaned and the number of these particles can be reduced. However, there are many types of this cleaning operation. It is effective in reducing the particles in the flow path quickly and with high certainty to implement a cleaning recipe consisting of one of the many types or a combination of multiple types according to the contaminated location (abnormal location).

[0023] In the resist coating apparatus 1, by using the foreign matter detection mechanism 3 described above, it is possible to perform a test (referred to as a residence test) for estimating the member that is the contaminated location, and to present to the user of the apparatus a cleaning recipe according to the result of the residence test. Then, the presented cleaning recipe is executed according to the user's instruction. The execution of the cleaning recipe is a countermeasure operation against the abnormality.

[0024] The above-mentioned residence test, presentation of the cleaning recipe, and execution of the cleaning recipe can be performed at the time of startup and during operation of the resist coating apparatus 1. The time of startup of the apparatus is the period before starting the processing on the wafer W, and more specifically, the period before flowing the resist through the piping system 2. Also, specifically, the time of operation of the apparatus is the period during which the wafers W are sequentially transported to the resist coating apparatus 1.

[0025] Regarding the residence test performed at the time of starting up the apparatus, it will be described with reference to the flow chart of FIG. 2. For convenience of explanation, regarding the flow path 20 from the tip of the nozzle 21 to the upstream end of the pipe 22, it is assumed that the volume from the tip of the nozzle 21 to the valve V1 is 20 mL, the volume from the valve V1 to the pump 23 is 30 mL, and the volume from the pump 23 to the upstream end of the pipe 22 is 100 mL. Since it is at the time of starting up the apparatus, a thinner storage section 28 and a flow path forming member 19 are attached to the piping system 2 instead of the resist storage section 25 and the filter 24.

[0026] In this residence test, by supplying N2 gas from the N2 gas supply source 27 to the thinner storage section 28, opening and closing the valves V, and operating the pump 23, the entire flow path 20 is filled with the thinner supplied from the thinner storage section 28 (step R1). After this filling, the operation of the pump 23 is stopped and the supply of N2 gas from the N2 gas supply source 27 is also stopped (that is, the valve V3 is in the closed state). Thereby, no flow is formed in the thinner in the flow path 20, and it remains in the flow path 20 (step R2). Note that the time point when the operation of the pump 23 is stopped and the valve V3 is closed is set as t1 so that such a state is formed. After the time point t1, foreign matter at the contaminated location is released as particles into the thinner filled and retained in the flow path 20. Since no flow of the thinner is formed, this particle remains near the contaminated location. Therefore, the number of particles at a local position in the flow path 20 increases. These steps R1 and R2 correspond to the residence process.

[0027] At a subsequent time point t2, the pump 23 performs a liquid suction and discharge operation, forming a flow of thinner from the thinner storage unit 28 toward the nozzle 21, and the thinner remaining in the flow path 20 is discharged from the nozzle 21. Thereby, the particles discharged from the contaminated portion into the flow path pass through the light transmission portion 31 and are discharged from the nozzle 21. For example, when discharging the volume of thinner in the flow path 20, the operation of the pump 23 is stopped, and the discharge of thinner from the nozzle 21 is terminated. This discharge end time point is set as t3. During the period from the above time point t2 to the time point t3, the discharge amount of thinner from the nozzle 21 is monitored, and the acquisition of the detection signal from the foreign matter detection mechanism 3 is continued (step R3). This step R3 corresponds to a flow-through process, a discharge amount detection process, and a signal acquisition process. Then, from the discharge amount of this thinner and the detection signal, the relationship between the discharge amount of the thinner and the number of particles having a particle size within the set range is obtained (step R4).

[0028] The relationship obtained in step R4 can be specifically represented as, for example, the graph shown in FIG. 3. The horizontal axis and the vertical axis of the graph in FIG. 3 show the discharge amount of thinner and the number of particles, respectively. When the number of particles at a local position in the flow path 20 increases due to the release of particles from the above-described contaminated portion, the waveform of the graph shows a peak at the discharge amount corresponding to that local position. The position of the nozzle 21 in the flow path 20 and the position of the light transmission portion 31 where particle detection is performed are slightly offset. Therefore, in the flow path 20, it can be estimated that the position upstream from the tip of the nozzle 21 toward the flow path 20 by the total amount of the volume (mL) from the tip of the nozzle 21 to the light transmission portion 31 and the discharge amount (mL) at which the peak appears is the contaminated portion. Since the number of particles is detected by detecting foreign matter from the detection signal from the above-described foreign matter detection mechanism 3, estimating the contaminated portion from the waveform of the graph in this way is based on the discharge amount when the particles are detected.

[0029] When the graph of FIG. 3 is obtained, since the peak of the waveform appears at a position where the discharge amount is slightly less than 50 mL, it can be estimated that the pump 23 located on the upstream side of the flow path 20 by a volume of 50 mL from the tip of the nozzle 21 is the contaminated part. Thus, since it is possible to estimate the contaminated part from the relationship between the discharge amount of the thinner and the number of particles, it is possible to present a cleaning recipe as described above. In FIG. 3, L1 and L2 are the first setting range and the second setting range for the discharge amount, respectively. The discharge amount in the first setting range L1 is the same amount as or approximately the same volume as the volume from the tip of the nozzle 21 to the pump 23, and the discharge amount in the second setting range L2 is the same amount as or approximately the same volume as the volume from the tip of the nozzle 21 to the valve V1. The presentation of the cleaning recipe is performed by determining whether the peak of the waveform of the above graph falls within the setting ranges L1 and L2 of these discharge amounts. By the way, hereinafter, the graph showing the relationship between the discharge amount of the liquid from the nozzle 21 and the number of particles obtained in the residence test as described above will be referred to as a discharge amount vs. number of particles graph.

[0030] Regarding the residence test during the operation of the apparatus, it is substantially the same as the residence test at the start-up of the apparatus. The differences are that when performing the residence test during wafer processing, a resist storage unit 25 is attached to the piping system 2, and the resist from the resist storage unit 25 is supplied to the flow path 20, so that the resist is retained in the flow path 20 instead of the thinner. Also, as described above, during the processing of the wafer W, a filter 24 is provided in the piping system 2 instead of the flow path forming member 19, so the difference is also that the members constituting the flow path 20 are different in this way.

[0031] By the way, in this residence test, regarding the time from the above-mentioned time t1 to time t2 (referred to as the residence time) for retaining the thinner or the resist, the particles are sufficiently discharged into the flow path 20 to improve the test accuracy, and the stop time of the processing of the wafer W is set so as not to be excessive. From that viewpoint, the residence time is preferably, for example, 0.5 hours to 12 hours, and more preferably, for example, 0.5 hours to 9 hours. And regarding the residence test during the operation of the apparatus, from the viewpoint of more surely preventing the drying and solidification of the resist in the flow path 20, it is more preferable to set this residence time to, for example, 2 hours or less.

[0032] To further explain the residence time, when the apparatus is in operation, the wafer W is conveyed to the resist coating apparatus 1 at regular intervals and processed. In sequentially processing the wafer W in such a manner, the liquid suction operation and the multiple discharge operations of the pump 23 are repeated. Therefore, the resist at the upstream end of the flow path 20 gradually moves to the nozzle 21 and is discharged onto the wafer W. If the time until the resist is discharged from the upstream end of the flow path 20 to the nozzle 21 is defined as X, for the purpose of sufficiently discharging the particles into the flow path 20, the above-mentioned residence time is set to be longer than the time X. Note that the time X is also the time from an arbitrary timing until the timing when the discharge amount per discharge by the pump 23 × the number of discharges becomes larger than the volume of the flow path 20.

[0033] Subsequently, five different types of examples of the cleaning operations constituting the cleaning recipe will be described. These five types of cleaning operations are referred to as line cleaning, gas-liquid replacement, pressure purge, pump degassing, and short dummy. Regarding these cleaning operations, since the thinner in the resist storage section 25 connected to the piping system 2 or the resist in the thinner storage section 28 is used as the cleaning liquid, the cleaning liquid used is different between when the apparatus is started up and when the apparatus is in operation. Here, the cleaning at the start-up of the apparatus will be described assuming that the thinner is the cleaning liquid.

[0034] For the five types of cleaning operations, it is common to flow the cleaning liquid from the upstream side to the downstream side in the flow path 20. Some of them discharge the cleaning liquid from the nozzle 21. For the line cleaning, the N2 gas from the gas supply source 27 is supplied for a preset period with the valves V1 to V3 open, so that the thinner in the thinner storage section 28 is pumped to the nozzle 21 and discharged from the nozzle 21 to clean the flow path 20.

[0035] For the gas-liquid replacement, first, the same operation as the line cleaning is performed to pump and discharge the thinner in the thinner storage section 28 to the nozzle 21. Even when the thinner in the thinner storage section 28 runs out, the supply of N2 gas to the thinner storage section 28 is continued, so that the thinner is removed from the flow path 20 and the N2 gas is discharged from the nozzle 21 to purge the flow path 20, and the thinner in the flow path 20 is replaced with N2 gas. Then, the operator replaces the thinner storage section 28 with one filled with thinner. And, by supplying N2 gas from the gas supply source 27, the flow path 20 is refilled with thinner. Such line cleaning, replacement of the flow path 20 with N2 gas, replacement of the thinner storage section 28, refill of the flow path 20 with thinner, and line cleaning are repeated a predetermined number of times.

[0036] For the pressure purge, the pump 23 and each of the valves V1 to V3 operate in the same way as when discharging the resist to the wafer W, so that the thinner is discharged from the nozzle 21. However, the discharge pressure of the pump 23 during the execution of this pressure purge is set higher than the discharge pressure when processing the wafer W. For the pump degassing, the valves V4 and V2 are open, the valves V1 and V3 are closed, and the pump 23 is operated to push out the bubbles accumulated in the pump by the cleaning liquid supplied from the upstream side of the pump 23 to the discharge pipe 18 for removal. For the short dummy, after the pump 23 sucks the liquid, the discharge operation of the pump 23 and the opening and closing of the valve V1 are repeated to discharge the thinner from the nozzle 21. The interval between the discharges of the thinner from this nozzle 21 is shorter than the discharge interval of the resist when sequentially processing the wafer W during the operation of the apparatus.

[0037] Among the cleaning operations described above, for the line cleaning and gas-liquid replacement, the entire flow path 20 can be effectively cleaned. Regarding the gas-liquid replacement among them, the entire flow path 20 can be particularly effectively cleaned at the time of starting up the apparatus. The pressure purge can effectively clean the region downstream of the pump 23 in the flow path 20. The pump degassing is an effective cleaning operation when the bubbles accumulated in the pump 23 become particles. The short dummy is an effective cleaning method when the pump 23 and / or the valve V1 become contaminated parts. Note that during the operation of the apparatus, a resist storage unit 25 is attached to the piping system 2 instead of the thinner storage unit 28, so that resist is used as the cleaning liquid instead of thinner. Except for such differences, each cleaning operation can be performed in the same manner during the operation of the apparatus and at the time of starting up the apparatus.

[0038] Note that the foreign matter detection mechanism 3 is also used to determine the adequacy of the cleaning of the flow path 20 by a cleaning recipe including the above-described cleaning operations. During the flow of the resist or thinner through the light transmission part 31 in the cleaning operation, the control unit 4 acquires the detection signal from the foreign matter detection mechanism 3, and determines whether or not the flow path 20 has been cleaned based on the number of particles obtained from this detection signal.

[0039] The control unit 4 will be described with reference to FIG. 1. As described above, the control unit 4 is a computer and includes a program 41, a notification unit 42, an operation unit 43, and a memory 44. The program 41 transmits control signals to each part of the resist coating apparatus 1 to control operations such as opening and closing of each valve V, operation of the pump 23, and light irradiation by the foreign matter detection mechanism 3, so that the above-described cleaning operation and processing on the wafer W are performed. And this program 41 is configured so that the residence flow described in FIG. 2 and the work flows shown in FIGS. 4 and 6 described later proceed. Therefore, the program 41 is also configured to be able to acquire a detection signal from the foreign matter detection mechanism 3, detect the discharge amount of the liquid from the nozzle 21, create a discharge amount vs. particle number graph from the detection signal and the discharge amount, specify the peak of the graph, compare the peak with the set range of the discharge amount, and present a cleaning recipe based on the comparison. The program 41 is stored and installed in a storage medium such as a compact disk, a hard disk, a memory card, a DVD, etc.

[0040] The notification unit 42 is constituted by, for example, a display or a speaker, and presents a cleaning recipe to the user of the apparatus by screen display or voice. Also, as will be described later, when the flow path 20 cannot be cleaned even if the cleaning recipe is repeated (when the startup operation ends abnormally and when the recovery of the apparatus is impossible), the notification unit 42 notifies the user to that effect. The operation unit 43 is constituted by buttons, switches, etc., and the user can execute the presented cleaning recipe by performing a predetermined operation on the operation unit 43. The memory 44 stores the capacity from the tip of the nozzle 21 to each member in the flow path 20 described above, parameters in each cleaning operation (execution time, number of discharge operations of the pump 23, etc.), the order of performing each cleaning operation constituting the cleaning recipe, and the first and second set ranges L1, L2, etc. for the discharge amount. That is, the memory 44 stores information necessary for executing the residence test and information necessary for presenting and executing the cleaning recipe.

[0041] Next, with reference to FIG. 4, an example of the flow of the operation to clean the flow path 20 at the time of starting up the apparatus (hereinafter referred to as the startup operation) will be described. First, when the thinner supplied from the thinner storage unit 28 is supplied to the flow path 20 by the supply of N2 gas from the N2 gas supply source 27 and the operation of the pump 23, line cleaning is performed (step S1). For example, during a predetermined length of inspection period near the end of this line cleaning, the number of particles is detected based on the detection signal from the foreign matter detection mechanism 3, and it is determined whether or not the number is less than or equal to a reference value (step S2).

[0042] If it is determined in step S2 that the number of particles is less than or equal to the reference value, it is assumed that the startup operation has ended normally. When it is assumed that the startup operation has ended normally in step S2 or in other steps S described later, the thinner storage unit 28 is replaced with the resist storage unit 25, and the flow path forming member 19 is replaced with the filter 24, respectively. Then, by the operation of the pump 23, the resist is made to flow from the resist storage unit 25 to the tip of the nozzle 21 so that the wafer W can be processed.

[0043] If it is determined in step S2 that the number of particles is not less than or equal to the reference value, line cleaning is performed again (step S3). Then, the number of particles is detected based on the detection signal during a predetermined inspection period in the same manner as in step S2, and it is determined whether or not the number is less than or equal to the reference value (step S4). In step S4, if it is determined that the number of particles is less than or equal to the reference value, it is assumed that the startup operation has ended normally. If it is determined in step S4 that the number of particles is not less than or equal to the reference value, the above-described residence test is performed (step S5), and the peak of the waveform of the discharge amount vs. particle number graph described in FIG. 3 and the discharge amount of the thinner corresponding to the peak are specified.

[0044] In the following description, for the sake of convenience, as a result of the above residence test, it is assumed that there is only one peak in the waveform of the graph or there is no distinct peak. If the discharge amount corresponding to the peak is within the first set range L1, a cleaning recipe A1 that is set to have high cleanability for the pump 23 is presented. If the discharge amount corresponding to the above peak is within the second set range L2, a cleaning recipe A2 that has high cleanability for the valve V1 is presented. If it is impossible to specify the discharge amount corresponding to the peak due to the absence of a distinct peak, or if a peak appears in a range outside the first set range L1 and the second set range L2, a cleaning recipe A3 for cleaning the entire flow path 20 is presented. By the way, since the peak of the graph corresponds to the contaminated part as described with reference to FIG. 3, the step of specifying the peak of the waveform corresponds to the step of estimating the contaminated part.

[0045] After any one of the cleaning recipes A1 to A3 is presented as described above, when the user of the apparatus instructs the start of the cleaning recipe, the presented cleaning recipe is executed. When the cleaning recipe A1 is executed, gas-liquid replacement (step S11), pump degassing (step S12), pressurized purge (step S13), and line cleaning (step S14) are executed in this order. For operations that require work by an operator, such as the replacement of the resist storage unit 25 in the gas-liquid replacement in step S11, each step proceeds in order when the operation is performed. The same applies to the other cleaning recipes described later.

[0046] Based on the detection signal from the foreign matter detection mechanism 3 obtained during the above-described inspection period during the line cleaning in step S14, the number of particles is detected. Then, it is determined whether or not the number is less than or equal to the reference value (step S15). If it is determined in step S15 that the number of particles is less than or equal to the reference value, it is assumed that the startup operation has been completed normally. If it is determined in step S15 that the number of particles is not less than or equal to the reference value, the cleaning recipe A1 (steps S11 to S14) and the determination of the number of particles associated therewith (step S15) are executed again.

[0047] In this way, the determination in step S15 of cleaning recipe A1 is repeated. For example, if it is determined that the number of particles is not less than the reference value even in step S15 after cleaning recipe A1 has been repeated n times (n is an integer), the repetition of cleaning recipe A1 and step S15 stops, and the startup operation is considered to have ended abnormally. Then, the operator takes measures such as replacing the pump 23.

[0048] When cleaning recipe A2 is executed, a short dummy (step S21) and line cleaning (step S22) are executed in this order. Then, during the inspection period near the end of the line cleaning in step S22, it is determined whether the number of particles is less than or equal to the reference value (step S23). If it is determined in step S23 that the number of particles is less than or equal to the reference value, the startup operation is considered to have ended normally. If it is determined that the number is not less than the reference value, cleaning recipe A2 (steps S21 to S22) and the accompanying determination (step S23) are executed again. The determination in step S23 of cleaning recipe A2 is repeated. If it is determined that the number of particles is not less than the reference value even in step S23 after cleaning recipe A2 has been repeated n times, the repetition of cleaning recipe A2 and step S23 stops, and the startup operation is considered to have ended abnormally. Then, the operator takes measures such as replacing the valve V1.

[0049] When the cleaning recipe A3 is executed, gas-liquid replacement (step S31), pump degassing (step S32), pressure purge (step S33), and line cleaning (step S34) are executed in this order. Then, during the inspection period near the end of the line cleaning in step S34, it is determined whether the number of particles is less than or equal to the reference value (step S35). Thus, the cleaning operations of the same type as those in the cleaning recipe A1 are performed in the same order in the cleaning recipe A3 (steps S31 to S34). And, similar to the case when the cleaning recipe A1 is executed, depending on the determination in step S35, it is decided whether the cleaning recipe A3 and the determination are repeated. If it is determined that the number of particles is less than or equal to the reference value in step S35 even after repeating n times, it is considered that the startup operation has ended abnormally. In such a case of abnormal termination, the operator performs optional measures such as testing and replacing each member of the piping system 2. Note that the identification of the peak of the graph and the execution of any one of the cleaning recipes A1 to A3 correspond to the corresponding steps performed based on the detection signal.

[0050] By the way, as a result of performing the residence test, there are cases where multiple peaks appear in the waveform of the graph of the discharge amount versus the number of particles. In that case, a cleaning recipe combining the cleaning recipes corresponding to each peak is presented. A specific example is shown for the case where peaks appear in the first setting range L1 and the second setting range L2 as shown in FIG. 5. A cleaning recipe including the cleaning operations (gas-liquid replacement, pump degassing, pressure purge, line cleaning) constituting the cleaning recipe A1 corresponding to the first setting range L1 and the cleaning recipe A4 (short dummy, line cleaning) corresponding to the second setting range L2 is presented. The order of the cleaning operations in the cleaning recipe is determined by, for example, a predetermined algorithm.

[0051] Figure 5 shows cleaning recipe A4, which is an example of a cleaning recipe presented as a combination of cleaning recipes A1 and A2. The cleaning recipe A4 is composed of gas-liquid replacement (step S41), pump degassing (step S42), pressurized purge (step S43), short dummy (step S44), and line cleaning (step S45), and these cleaning operations are executed in sequence. Similar to cleaning recipes A1 to A3, a determination (step S46) is made regarding the number of particles during line cleaning. Then, depending on the result of this determination, it is decided whether cleaning recipe A4 (steps S41 to S45) and the determination (step S46) are repeated, or the startup operation is considered to have ended normally. If it is determined that the number of particles is not below the reference value in step S46 even after cleaning recipe A4 is repeated n times, the startup operation is considered to have ended abnormally.

[0052] When combining the above cleaning recipes, for the purpose of efficiently cleaning each contaminated area and avoiding excessive cleaning, based on the height of each peak of the graph, the parameters contributing to the cleaning power in specific cleaning operations may be changed before and after the combination. As an example, for the parameters of the cleaning operation specific to cleaning recipe A1 corresponding to the first setting range L1, in cleaning recipe A4, they are changed to be the height of the peak of the first setting range L1 / the sum of the heights of the first and second peaks. The parameters of the cleaning operation specific to cleaning recipe A2 corresponding to the second setting range L2 are changed in the same way.

[0053] Describing the above example in more detail, assume that the height of the peak of the first setting range L1: the height of the peak of the second setting range = 1:2 as illustrated in FIG. 5. Among the cleaning operations constituting the cleaning recipes A1 and A2, for the pressure purge, which is a cleaning operation unique to the cleaning recipe A1, assume that in step S13 of the cleaning recipe A1, the discharge operation of the pump 23 is set to be performed D times. In that case, for example, for the pressure purge in step S43 of the cleaning recipe A4, the discharge operation is performed D times × 1 / (1 + 2), for example. Similarly, for the short dummy, which is a cleaning operation unique to the cleaning recipe A2 among the cleaning recipes A1 and A2, assume that in the cleaning recipe A2, the discharge operation of the pump 23 and the opening / closing operation of the valve V1 are set to be performed E times. In that case, for example, in the cleaning recipe A4, the operations are performed E times × 2 / (1 + 2), for example.

[0054] In the combination, although an example was shown in which the parameters of the cleaning operations unique to the cleaning recipes A1 and A2 are changed, it may be an arbitrary setting as to which cleaning operation parameters are changed. For example, as the parameter, for gas-liquid replacement, the number of replacements may be changed, for line cleaning, the time for flowing thinner through the flow path 20 may be changed, and for pump degassing, the number of discharge operations of the pump 23 may be changed.

[0055] Subsequently, regarding the operation flow for cleaning the flow path 20 during the operation of the apparatus, the difference from the flow at the start-up of the apparatus will be mainly described with reference to FIG. 6. For the wafer W sequentially conveyed to the resist coating apparatus 1, resist is discharged from the nozzle 21 to form a resist film. During the discharge of the resist, a detection signal output from the foreign matter detection mechanism 3 is acquired, and the number of particles is detected. If it is determined that this number is not less than the reference value, the loading of a new wafer W into the resist coating apparatus 1 is temporarily stopped.

[0056] After the transfer is stopped, line cleaning is performed (step T1). Based on the detection signal obtained during the inspection period near the end of this line cleaning, the number of particles is detected, and it is determined whether the number is less than or equal to the reference value (step T2). If it is determined in step T2 that the number of particles is less than or equal to the reference value, the recovery of the apparatus is considered successful. When recovery is successful in step T2 or in other step T described later, the transfer and processing of the wafer W to the resist coater 1 are restarted.

[0057] If it is determined in step T2 that the number of particles is not less than or equal to the reference value, a residence test is performed, and a graph of the discharge amount from the nozzle 21 versus the number of particles is obtained (step T3). Then, the peak of the waveform and the discharge amount corresponding to the peak are specified, and a cleaning recipe is presented according to the specified peak.

[0058] Here too, for the sake of convenience of explanation, it is assumed that there is only one peak in the waveform of the graph or that there is no distinct peak. If the discharge amount at which the peak appears is within the first set range L1, a cleaning recipe B1 with high cleanability for the pump 23 is presented. If the discharge amount at which the peak appears is within the second set range L2, a cleaning recipe A2 with high cleanability for the valve V1 is presented. If no distinct peak exists and the discharge amount corresponding to the peak cannot be specified, or if the peak appears in a range outside the first set range L1 and the second set range L2, it is assumed that automatic recovery of the apparatus is impossible and no cleaning recipe is presented. That is, a determination is made as to whether the problem can be addressed with a cleaning recipe (step T4). When no cleaning recipe is presented, for example, an operator removes each member in the apparatus and performs tests to identify the cause.

[0059] After any one of cleaning recipes B1 to B2 is presented and the user instructs the start of the cleaning recipe, the presented cleaning recipe is executed. When cleaning recipe B1 is executed, line cleaning (step T11) and pressurized purge (step T12) are executed in this order. In the pressurized purge of step T12, the discharge operation of pump 23 is performed the set number of times. For example From the detection signals in each previous time by a predetermined number of times from the last time of the discharge operation, the number of particles is detected. Then, it is determined whether or not this number is less than or equal to the reference value (step T13).

[0060] If it is determined in step T13 that the number of particles is less than or equal to the reference value, it is considered that the recovery of the apparatus has been successful. If it is determined in step T13 that the number of particles is not less than or equal to the reference value, cleaning recipe B1 (steps T11 to T12) and the accompanying determination (step T13) are executed again. In this way, the determination of cleaning recipe B1 and step T13 is repeated. If it is determined in step T13 even after cleaning recipe B1 has been repeated m times (m is an integer) that the number of particles is not less than or equal to the reference value, the repetition of cleaning recipe B1 and the determination of step T13 stops, and it is assumed that automatic recovery of the apparatus is impossible, and a predetermined test is performed by an operator.

[0061] When the cleaning recipe B2 is executed, a short dummy (step T21) and line cleaning (step T22) are executed in this order. During the line cleaning in step T22, detection signals are acquired in the same manner as when the line cleaning in step T2 is executed, and the number of particles is detected from the detection signals during the inspection period near the end of the line cleaning. Then, it is determined whether the number of the particles is less than or equal to a reference value (step T23). If it is determined in step T23 that the number of particles is less than or equal to the reference value, it is considered that the recovery of the device has been successful. If it is determined in step T23 that the number of particles is not less than or equal to the reference value, the cleaning recipe B2 (steps T21 to T22) and the accompanying determination (step T23) are executed again. In this way, the cleaning recipe B2 and the determination in step T23 are repeated. If it is determined in step T23 even after the cleaning recipe B2 has been repeated m times (m is an integer) that the number of particles is not less than or equal to the reference value, the repetition of the cleaning recipe B2 and the determination in step T23 stops, and it is considered that automatic recovery of the device is impossible, and a predetermined test is performed by an operator.

[0062] Regarding the cleaning recipes B1 and B2 presented with the pump 23 and the valve V1 as the contaminated parts respectively, the cleaning operations included are fewer and simpler than those of the cleaning recipes A1 and A2 at the start-up of the device presented with the pump 23 and the valve V1 as the contaminated parts respectively, for the purpose of quickly recovering the device and improving the productivity of the device. For the same purpose, when it is impossible to specify the peak or when it appears in a range other than the first and second setting ranges L1 and L2, no cleaning recipe is presented and automatic recovery is considered impossible. Also, the cleaning recipes A1 to A3 and B1 to B3 can be repeated. However, for the above purpose, regarding the relationship between the number of times n that the cleaning recipes A1 to A3 can be repeated shown in each of the above-mentioned flows and the number of times m that the cleaning recipes B1 to B2 can be repeated, for example, n > m. As an example, n is 5 and m is 1.

[0063] Note that it is conceivable that this filter 24 may become a contaminated part during the operation of the apparatus. Therefore, a third setting range L3 regarding the discharge amount corresponding to the position of the filter 24 and a cleaning recipe B3 with high cleanability for the filter 24 are set. When the peak of the graph in the residence test appears within the third setting range L3, this cleaning recipe B3 may be presented and executed. Also, for each cleaning recipe during this wafer processing, similar to the cleaning recipe at the startup of the apparatus, when multiple peaks appear in the residence test, they can be combined.

[0064] According to the resist coating apparatus 1 configured as described above, contamination locations are estimated at the startup and during the operation of the apparatus, and cleaning recipes corresponding to those contamination locations are presented. Therefore, the cleaning of the contamination locations can be performed efficiently, so that the wasteful use of thinner and resist as cleaning liquids can be suppressed, and the state in which the apparatus becomes unusable for the wafer W can be prevented from continuing for a long time. As a result, the productivity of semiconductor products can be increased. Also, since the frequency with which an operator removes each member from the piping system 2 to individually verify whether it is a contamination location in order to identify the contamination location is reduced, the burden on the operator is alleviated.

[0065] Suppose that without using the foreign matter detection mechanism 3, after supplying a predetermined amount of cleaning liquid to the flow path, the cleaning liquid is discharged from the nozzle 21 onto the wafer W on the spin chuck 12, and the wafer W is transported to a detection device outside the resist coating apparatus 1 to detect particles. In that case, it takes time to transport the wafer W outside the resist coating apparatus 1. And by performing the detection of the particles, the wafer W will be used without being processed. Therefore, the resist coating apparatus 1 also has the effect of preventing such labor for carrying out the wafer W and preventing the wafer W from being subjected to processing that does not contribute to the production of semiconductor devices.

[0066] Incidentally, the configuration of each cleaning recipe described above is an example, and the types and order of cleaning operations included in the recipe can be appropriately changed. Also, although each cleaning recipe is configured by a plurality of cleaning operations, it may perform only one cleaning operation. Further, when the selection and combination of the above-described cleaning recipes are performed by the residence test, the control unit 4 may be configured so that the cleaning recipe is automatically executed regardless of the user's instruction. Therefore, a device configuration in which the cleaning recipe is not presented to the user may be used.

[0067] Also, among the members forming the flow path 20, the valve V1, the pump 23, and the filter 24 are exemplified as the members that become the contamination sites. However, as other members that can become the contamination sites, a cleaning recipe may be set, and the cleaning recipe may be presented and executed according to the result of the residence test. For example, although the description is omitted, a tank for temporarily storing the resist is provided between the filter 24 and the resist storage unit 25. The tank may be presumed to be a contamination site, and a cleaning recipe with high cleanability for the tank may be prepared.

[0068] Then, the control unit 4 may notify the user by, for example, displaying on the display forming the notification unit 42 a discharge amount vs. particle number graph obtained by the residence test. The user who views the graph may replace the member corresponding to the peak position of the waveform. That is, a device configuration in which the cleaning recipe is not executed may be used. Also, in this case, it is the user rather than the control unit 4 who estimates the contamination site. Therefore, the estimation of the contamination site is not necessarily limited to the control unit 4.

[0069] Next, a modification of the above-described embodiment will be described. In the embodiment described above, regarding the line cleaning in the cleaning recipe A1 that is performed once or repeatedly, the time from the start to the end (referred to as the line cleaning time) is a fixed time set in advance. In this modification, the control unit 4 determines the line cleaning time of the last cleaning recipe A1 among the cleaning recipes A1 that are performed once or repeatedly in this way, based on the result of the residence test. In order to make such a determination possible, machine learning is performed on the control unit 4.

[0070] The above machine learning will be described. As described in the above embodiment, when the flow at the startup of the apparatus shown in FIG. 4 progresses and the residence test in step S4 is performed, assuming that the discharge amount vs. particle number graph obtained as a result presents the cleaning recipe A1. The control unit 4 stores the discharge amount vs. particle number graph in the memory 44. Then, the execution of the cleaning recipe A1 is instructed, and the steps S11 to S15 described above are executed. During the execution of the line cleaning in step S14 (that is, from the start time to the end time of the line cleaning), the acquisition of the detection signal from the foreign matter detection mechanism 3 continues. This line cleaning time is a fixed time as described above. The determination in step S15 is made based on what is obtained during the inspection period near the end of the line cleaning, in the same manner as in the example described above. That is, a part of the detection signals obtained during the line cleaning time is used for the determination.

[0071] Assume that the cleaning recipe A1 and the determination in step S15 are performed in this way, and the startup operation has been completed normally. As described in the flowchart of FIG. 4, the cleaning recipe A1 is performed one or more times by the time the startup operation is completed. During the line cleaning period of the last cleaning recipe A1 among them, the detection signal continuously obtained and the number of repetitions of the cleaning recipe A1 are associated with the discharge amount vs. particle number graph obtained immediately before the execution of the cleaning recipe A1 and stored in the memory 44. The control unit 4 specifies the convergence point at which the number of particles does not exceed the reference value from the stored detection signal. That is, between this convergence point and the end point of line cleaning, the number of particles does not exceed the reference value. Therefore, specifying this convergence point is to detect the minimum line cleaning time required to ensure that the number of particles does not exceed the reference value for the last line cleaning among the one or more line cleanings performed. The above machine learning is performed each time the cleaning recipe A1 is presented and executed by the residence test.

[0072] After that, the result of the learning is output. Specifically, when a discharge amount vs. particle number graph presenting the cleaning recipe A1 is obtained by performing the residence test, the graph with the highest waveform similarity (similarity) is specified from among the graphs stored in the memory 44. Then, for example, the cleaning recipe A1 is repeated the number of times of repetition stored corresponding to the specified graph. For the last line cleaning among the one or more line cleanings performed, it is determined to perform line cleaning until the convergence point corresponding to that graph. That is, the line cleaning time in line cleanings other than the last one is the time preset in the same way as in the learning stage, but for the last line cleaning, the end point is advanced. In accordance with advancing the end point, the inspection period for making the determination in step S15 in the last line cleaning is also advanced. And if it is determined that it is not below the reference value in step S15, it is assumed that the startup operation has ended abnormally because particles cannot be reduced due to unexpected factors.

[0073] By using the learning results as described above, it is possible to prevent the situation where, despite the fact that sufficient cleaning of the flow path 20 is impossible due to unexpected factors, line cleaning is performed in vain, resulting in the unnecessary use of thinner and time. Note that as output, it is not limited to simply applying the results at the convergence point obtained in the past and executing the cleaning recipe A1. Instead, according to a predetermined algorithm, the similarity between the graph obtained at the output stage after the learning stage and the graph obtained at the learning stage can be calculated, and the convergence point can be changed according to the similarity. For cleaning recipes A2, A3, and B2 as well, the line cleaning time can be changed from the originally set time by machine learning in the same manner as cleaning recipe A1.

[0074] Regarding cleaning recipe B1, unlike cleaning recipe A1, the last cleaning operation of the recipe is a pressure purge. However, machine learning is performed in substantially the same manner as cleaning recipe A1, and the number of discharge operations of pump 23 performed during the pressure purge can be changed from the preset number of times. Focusing on the differences between the learning and output for cleaning recipe A1, in the learning stage, when the cleaning recipe B1 is executed and the pressure purge at step T12 is performed as a result of the residence test, a detection signal is acquired each time the discharge operation of pump 23 is performed. Therefore, in this learning stage, detection signals for the preset number of discharge operations are acquired. Then, for the last pressure purge among the one or more executions of cleaning recipe B1, the detection signals during the discharge operation of each pump are associated with the number of repetitions of cleaning recipe B1 and the discharge amount vs. particle number graph and stored in memory 44. Then, the control unit 4 determines the discharge number (referred to as the convergence number) at which the number of particles no longer exceeds the reference value in the discharge operation after which time during the last pressure purge.

[0075] As an output, when a residence test is performed and a discharge amount vs. particle number graph presenting cleaning recipe A1 is obtained, among the graphs obtained in the past when cleaning recipe B1 was executed, the graph with the highest waveform similarity is specified. Clean recipe B1 is repeated the number of times corresponding to that graph, and in the pressure purge in the last cleaning recipe B1, the discharge operation is performed only the number of convergence times, and step T13 is performed while performing the discharge operation the number of convergence times, and it is determined whether or not the device has been restored. That is, the discharge operation of the last cleaning recipe B1 is performed a number of times less than the preset number of times.

[0076] By the above machine learning and output, it is possible to prevent the pressure purge from being performed uselessly even though sufficient cleaning of the flow path 20 is impossible due to unexpected factors. Note that the number of convergence times obtained in the learning stage is not limited to being directly applied in the output stage for this cleaning recipe B1, and it may be changed according to the similarity of the graph, for example, in the same manner as cleaning recipe A1.

[0077] As described above, by machine learning, parameters such as the flow time of the cleaning liquid and the number of times of the discharge operation of the cleaning liquid by the pump 23, which affect the cleaning action of the flow path 20 in the cleaning operation, are set to be appropriate. This means that the appropriate timing to stop the cleaning operation is determined.

[0078] A further modification will be described. In this modification, machine learning is used to select a cleaning recipe after the residence test is executed. First, as described in each of the above embodiments, a residence test is performed to obtain a graph of discharge amount vs. particle number. The operator sets and executes an arbitrary cleaning recipe. In the last cleaning operation that constitutes the cleaning recipe, the number of particles is determined in the same manner as in the above-described example. If it is determined that the number of particles is equal to or less than the reference value, the graph of discharge amount vs. the number of particles and the content of the executed cleaning recipe are associated and stored in the memory 44. This is repeated for learning. That is, an effective cleaning recipe is repeatedly learned for the waveform of the graph.

[0079] As the output of the learning, when a discharge amount vs. particle number graph is obtained by the residence test, the graph with the highest similarity is identified from the waveforms of the graphs obtained so far for the waveform of that graph. Then, the cleaning recipe corresponding to the identified graph is presented to or executed for the user. Therefore, in this example, a plurality of candidates are stored in the memory 44 as cleaning recipes, and when the residence test is performed, the one with the highest effectiveness in cleaning the flow path 20 is selected from among the candidates according to the result of the residence test and presented or executed.

[0080] Note that after performing the residence test to obtain a graph of the discharge amount vs. the number of particles, for example, an operator removes each member of the flow path 20 from the piping system 2 and performs an arbitrary test individually to identify the contaminated location. Learning may be performed to associate the identified contaminated location with the graph and store it in the memory 44. As the output, when a graph of the discharge amount vs. the number of particles is obtained, the graph with the highest similarity is identified from the graphs obtained in the learning stage. The contaminated location corresponding to the identified graph may be presented to the user as the estimated contaminated location. Therefore, in this example, a plurality of candidates are stored in the memory 44 as contaminated locations, and when the residence test is performed, the one with the highest possibility of being the contaminated location is selected from among the candidates according to the result of the residence test and presented.

[0081] Incidentally, regarding the machine learning for estimating such contaminated locations, data may be accumulated in the memory 44 for the graph of the discharge amount versus the number of particles obtained through the residence test and the identified contaminated locations, and a neural network may be constructed. As the neural network, for example, regarding the graph of the discharge amount versus the number of particles obtained by the residence test, the number of particles at predetermined discharge amount intervals may be input to the input layer. Therefore, the number of nodes in the input layer is provided corresponding to that interval. As the nodes in the output layer, for example, three nodes are provided, and the probability that the valve V1, the pump 23, and other members are contaminated locations is output from these nodes. Specifically, when the graph of the discharge amount versus the number of particles is obtained after the learning stage, the probability Y1% that the valve V1 is a contaminated location, the probability Y2% that the pump 23 is a contaminated location, and the probability Y3% that other members are contaminated locations may be displayed on the notification unit 42 in this manner.

[0082] Data may be similarly accumulated for the graph of the discharge amount versus the number of particles and the effective cleaning recipe to construct a neural network, and the effectiveness Z1% of the cleaning recipe A1, the effectiveness Z2% of the cleaning recipe A2,... may be displayed from the output layer of the network. Note that even when information corresponding to a rank (in the above case, the probability of being a contaminated location and the probability of effectiveness) is shown for such contaminated locations or cleaning recipes, it corresponds to the estimation of contaminated locations and the presentation of cleaning recipes. Also, once the effectiveness of the cleaning recipe is calculated in this way, the cleaning recipe with the highest effectiveness may be automatically executed.

[0083] Incidentally, when the control unit 4 detects the discharge amounts of thinner and resist from the nozzle 21, it is not limited to monitoring the operation of the pump 23. A flow meter may be provided in the pipe 22, and the detection may be performed based on the detection signal from the flow meter. Therefore, the discharge of the liquid retained in the above retention test may be performed by supplying N2 gas from the N2 gas supply source 27 to the resist storage unit 25 or the thinner storage unit 28. Further, in the above-described example, the liquid is discharged from the nozzle 21 when discharging the liquid, but it is not limited to discharging from the nozzle 21. A discharge path branched from the flow path 20 is provided on the downstream side of the flow path 20, and by opening and closing valves appropriately provided in the flow path 20 and the discharge path, the thinner and resist are discharged to the discharge path so that the flow in the flow path 20 is made possible.

[0084] Furthermore, the light receiving unit 33 of the foreign matter detection mechanism 3 is not limited to receiving side scattered light, and for example, it may be configured to receive forward scattered light. Also, the light received by the light receiving unit 33 from the flow path is not limited to scattered light, and the transmitted light in the light transmission unit 31 may be received.

[0085] The processing liquid supplied for processing the substrate is not limited to resist. For example, before supplying the resist, a thinner is applied as a processing liquid to the wafer W to modify the surface. A retention test may be executed, a cleaning recipe may be presented and executed for the flow path of the thinner. Other processing liquids may be a developer, a cleaning liquid for cleaning the wafer W, a chemical solution for forming an antireflection film, a chemical solution for forming an insulating film, an adhesive for bonding the wafer W, and the like. Also, the substrate to be processed is not limited to the circular wafer W, and may be, for example, a rectangular substrate for manufacturing a flat panel display or the like.

[0086] The embodiments disclosed this time should be considered as illustrative in all respects and not restrictive. The above embodiments may be omitted, substituted, changed, and combined in various forms without departing from the scope and spirit of the appended claims.

Explanation of Symbols

[0087] W Wafer 1 Resist Coating Device 21 Nozzle 22 Pipe 28 Resist Storage Section 32 Light Irradiation Section 33 Light Receiving Section

Claims

1. A step of flowing the processing liquid through a flow path connecting a storage unit for storing the processing liquid and a processing liquid supply unit, discharging the processing liquid from the processing liquid supply unit onto a substrate, and processing the substrate; A retention step of filling the flow path with a liquid and retaining it; A flowing step of flowing the retained liquid through the flow path; A signal acquisition step of irradiating the flow path with light by a light irradiation unit and receiving the light from the flow path by a light receiving unit while performing the flowing step, and acquiring a detection signal corresponding to foreign matter in the liquid output from the light receiving unit; A corresponding step of estimating an abnormal location in the flow path, presenting a countermeasure operation against the abnormality, or performing the countermeasure operation based on the detection signal; A liquid processing method comprising the above steps.

2. The liquid processing method according to claim 1, wherein in the retention step, the time for retaining the liquid is 0.5 hours or more.

3. Including a discharge amount detection step of detecting the amount of the liquid discharged from the processing liquid supply unit while performing the flowing step, The corresponding step of the liquid processing method according to claim 1 or 2 is performed based on the discharge amount when foreign matter is detected.

4. The corresponding step of the liquid processing method according to any one of claims 1 to 3 is performed based on machine learning performed until the corresponding step and the detection signal.

5. The corresponding step includes a step of presenting or performing the countermeasure operation, The countermeasure operation includes a cleaning operation of flowing a cleaning liquid through the flow path, The machine learning according to claim 4 is learning for determining the timing to end the cleaning operation.

6. The countermeasure operation includes a cleaning operation of flowing a cleaning liquid through the flow path, The corresponding step includes a step of presenting or performing a countermeasure operation selected from a plurality of candidates for the countermeasure operation, The machine learning according to claim 4 is learning for performing the selection of the countermeasure operation.

7. The corresponding step is as follows: A step of estimating an abnormal location in the flow path based on the detection signal and the amount of the liquid discharged from the processing liquid supply unit in the flowing step; A step of presenting or executing the countermeasure operation corresponding to the estimation result; The liquid processing method according to any one of claims 1 to 3, including the above steps.

8. A processing liquid supply unit for discharging and processing a processing liquid onto a substrate; A flow path connecting a storage unit for storing the processing liquid and the processing liquid supply unit; A flow mechanism for flowing a liquid through the flow path; A light irradiation unit for irradiating light onto the flow path; A light receiving unit for receiving light from the flow path; A step of flowing the processing liquid through the flow path, discharging it from the processing liquid supply unit onto the substrate, and processing the substrate; A retention step of filling and retaining a liquid in the flow path; A flow step of flowing the retained liquid through the flow path; A signal acquisition step of irradiating light onto the flow path by the light irradiation unit and receiving light from the flow path by the light receiving unit while performing the flow step, and acquiring a detection signal corresponding to foreign matter in the liquid output from the light receiving unit; A control unit that outputs a control signal so as to perform a correspondence step of estimating an abnormal location in the flow path, presenting a countermeasure operation against the abnormality, or implementing the countermeasure operation based on the detection signal; A liquid processing apparatus comprising the above.

9. The liquid processing apparatus according to claim 8, wherein the time for retaining the liquid in the retention step is 0.5 hours or more.

10. A discharge amount measurement mechanism for measuring the amount of the liquid discharged from the processing liquid supply unit in the flow step is provided, The control unit The correspondence step is performed based on the discharge amount when foreign matter is detected, according to the liquid processing apparatus of claim 8 or 9.

11. The correspondence step is performed based on machine learning performed until the correspondence step is performed and the detection signal, according to the liquid processing apparatus according to any one of claims 8 to 10.

12. The correspondence step includes a step of presenting or implementing the countermeasure operation, The countermeasure operation includes a cleaning operation of flowing a cleaning liquid through the flow path, The machine learning is learning for determining the timing to end the cleaning operation, according to the liquid processing apparatus of claim 11.

13. The countermeasure operation includes a cleaning operation of flowing a cleaning liquid through the flow path, The correspondence step includes a step of presenting or implementing a countermeasure operation selected from a plurality of candidates for the countermeasure operation, The machine learning is learning for performing the selection of the countermeasure operation, according to the liquid processing apparatus of claim 11.

14. The correspondence step is A step of estimating an abnormal location in the flow path based on the detection signal and the amount of the liquid discharged from the processing liquid supply unit in the flow process; A step of presenting or executing the countermeasure operation corresponding to the estimation result; The liquid processing apparatus according to any one of claims 8 to 10, which includes

15. A processing liquid supply unit for discharging and processing a processing liquid onto a substrate, A flow path connecting the storage unit in which the processing liquid is stored and the processing liquid supply unit, A flow mechanism for flowing a liquid through the flow path, A light irradiation unit for irradiating light onto the flow path, A light receiving unit for receiving light from the flow path, in a program used for a liquid processing apparatus including A step of flowing the processing liquid through the flow path, discharging it from the processing liquid supply unit onto the substrate, and processing the substrate, A retention step of filling the flow path with a liquid and retaining it, A flow step of flowing the retained liquid through the flow path, While performing the flow step, irradiating light onto the flow path by the light irradiation unit and receiving light from the flow path by the light receiving unit, and obtaining a detection signal corresponding to foreign matter in the liquid output from the light receiving unit, a signal acquisition step, Based on the detection signal, a corresponding step of estimating an abnormal location in the flow path, presenting a countermeasure operation for the abnormality, or implementing the countermeasure operation, A program configured to execute.

Images