JP7744101B2 - Droplet ejection device, parameter calculation method, and substrate processing system - Google Patents

Description

This disclosure relates to a droplet ejection device, a parameter calculation method, and a substrate processing system.

Droplet ejection devices are known that eject droplets to form images or two-dimensional or three-dimensional structures. Droplet ejection devices form objects by repeatedly ejecting droplets from nozzles, but various abnormalities can occur, such as the landing position of droplets ejected from these nozzles (hereinafter simply referred to as nozzle abnormalities).

Technology for detecting nozzle abnormalities has been devised (see, for example, Patent Document 1). Patent Document 1 discloses a droplet ejection device that includes a head in which nozzles that eject droplets are arranged, a detection unit that detects nozzle ejection abnormalities, and determines whether maintenance is required for the abnormal nozzle based on positional information within the head of the abnormal nozzle in which an ejection abnormality has been detected.

Japanese Patent Application Laid-Open No. 2017-177423

The purpose of this disclosure is to provide technology for early detection of nozzle abnormalities.

The present disclosure relates to a droplet ejection device that ejects droplets from a nozzle, comprising: an acquisition unit that acquires information related to the droplet ejection state; a time series data calculation unit that calculates time series data of the information related to the droplet ejection state; a calculation unit that calculates parameters related to cleaning of the nozzle based on the time series data; and a judgment unit that judges the timing of cleaning the nozzle based on the parameters , wherein the information related to the droplet ejection state is a droplet flight image, and the calculation unit calculates feature amounts detected from the droplet flight image, and calculates the parameters based on the time series data of the feature amounts .

This disclosure provides technology that allows for early detection of nozzle abnormalities.

1 is a cross-sectional view showing a main part of an organic EL display according to an embodiment. 1 is a plan view illustrating a substrate processing system according to an embodiment. FIG. 2 is a configuration diagram of a control device according to an embodiment. FIG. 1 is a plan view showing a coating apparatus according to an embodiment. FIG. 10 is a side view showing a state when the substrate is yawing by the left substrate moving unit. FIG. 2 is a plan view showing an arrangement of ejection heads according to an embodiment. FIG. 10 is a diagram illustrating an example of a method for capturing a droplet flying image. FIG. 3 is a functional block diagram illustrating an example of a cleaning function of the control device. FIG. 10 is a diagram showing an example of time-series data obtained from the landing positions of droplets ejected from normal nozzles. FIG. 10 is a diagram showing an example of time-series data obtained from the landing position of an abnormal nozzle. FIG. 10 is a diagram showing an example graph illustrating changes in impact positions over time. 10A and 10B are diagrams showing an example of droplet flight images in which different droplets ejected from the same nozzle are captured. FIG. 10 is a diagram showing an example of time-series data using curvature, which is a feature quantity detected from a droplet flight image. FIG. 10 is a diagram showing an example of droplet flight images ejected from several nozzles. FIG. 10 shows an example droplet flight image including droplet spots. FIG. 10 is a diagram showing an example droplet flight image including contamination of the nozzle plate. FIG. 10 is a diagram showing an example of a droplet flight image including a nozzle missing. 10A and 10B are diagrams showing an example of droplet flight images illustrating the difference in landing time between nozzles. FIG. 10 is a diagram illustrating an example of a method for determining an abnormal mode using time-series data. FIG. 10 is a diagram illustrating an example of a method for determining whether classification into clusters has been achieved. FIG. 10 is a flowchart illustrating an example of a procedure in which the control device monitors nozzle abnormalities. FIG. 10 is a flowchart illustrating an example of a procedure in which the control device monitors nozzle abnormalities. FIG. 10 is a diagram showing an example of a nozzle monitoring screen displayed on an operation panel connected to the control device.

Non-limiting exemplary embodiments of the present disclosure will be described below with reference to the accompanying drawings. As an example of a form for implementing the present disclosure, a method for calculating parameters related to nozzle cleaning will be described with reference to the drawings.

[Outline of the method for calculating the degree of abnormality]
In the present disclosure, the degree of abnormality of the nozzle that ejects droplets is calculated, and this degree of abnormality is output (displayed on a display) or used to determine whether cleaning should be performed. The degree of abnormality is calculated from the following two pieces of data:
(i) Time series data of impact locations
(ii) Time-series data of features acquired from droplet flight images (images of flying droplets) By focusing on time-series data, it becomes easier to detect abnormalities early (at the symptom stage). Furthermore, abnormalities can be detected earlier than when judged by humans.

In addition, the droplet ejection device can distinguish abnormal modes using time-series data on landing positions, time-series data on droplet flight images, or image analysis of droplet flight images. Appropriate cleaning can be performed depending on the abnormal mode.

The droplet ejection device disclosed herein therefore determines the timing of cleaning based on the degree of abnormality, allowing for the optimization of cleaning frequency. Furthermore, by optimizing cleaning frequency and performing cleaning according to the abnormality mode, the amount of material consumed during cleaning can be reduced. For example, materials used for substrates of liquid crystal displays using organic electroluminescence (EL) are expensive, but the amount of material consumed during cleaning can be minimized. Furthermore, if signs of abnormality can be detected, it becomes easier to schedule processes for the droplet ejection device, and cleaning can be performed before an abnormality occurs, reducing the need to redo the formation of target objects.

[Terminology]
A nozzle abnormality refers to some abnormality that differs from a normal state in the droplets ejected from the nozzle.

Nozzle cleaning parameters are information used to determine whether or when cleaning should be performed. In this embodiment, these are referred to as the degree of abnormality.

The information regarding the droplet ejection state can be any information that can be used to calculate this parameter. In this embodiment, we will use at least one of the impact position and droplet flight image as an example.

[Configuration of droplet ejection device]
Before describing the configuration of the device, the configuration of an organic EL display produced by the droplet ejection device will be described.

Figure 1 is a cross-sectional view showing the main parts of an organic EL display according to one embodiment. A transparent substrate such as a glass substrate or a resin substrate is used as the substrate 10. A TFT layer 12 is formed on the substrate 10. A planarization layer 18 is formed on the TFT layer 12 to flatten any steps formed by the TFT layer 12.

The planarization layer 18 is insulating. Contact plugs 19 are formed in contact holes that penetrate the planarization layer 18. The contact plugs 19 electrically connect the anodes 21, which serve as pixel electrodes formed on the flat surface of the planarization layer 18, to the TFT layer 12. The contact plugs 19 may be made of the same material as the anodes 21 and formed at the same time.

The organic light-emitting diode 13 is formed on the flat surface of the planarization layer 18. The organic light-emitting diode 13 has an anode 21 as a pixel electrode, a cathode 22 as a counter electrode provided on the opposite side of the pixel electrode from the substrate 10, and an organic layer 23 formed between the anode 21 and cathode 22. By operating the TFT layer 12, a voltage is applied between the anode 21 and cathode 22, causing the organic layer 23 to emit light.

The anode 21 is made of, for example, ITO (Indium Tin Oxide) and transmits light from the organic layer 23. Light that passes through the anode 21 passes through the substrate 10 and is extracted to the outside. An anode 21 is provided for each unit circuit 11.

The cathode 22 is made of, for example, aluminum, and reflects light from the organic layer 23 back toward the organic layer 23. The light reflected by the cathode 22 passes through the organic layer 23, the anode 21, and the substrate 10, and is extracted to the outside. The cathode 22 is common to multiple unit circuits 11.

The organic layer 23 includes, for example, a hole injection layer 24, a hole transport layer 25, an emissive layer 26, an electron transport layer 27, and an electron injection layer 28, in this order from the anode 21 side to the cathode 22 side. When a voltage is applied between the anode 21 and the cathode 22, holes are injected from the anode 21 into the hole injection layer 24, and electrons are injected from the cathode 22 into the electron injection layer 28. The holes injected into the hole injection layer 24 are transported by the hole transport layer 25 to the emissive layer 26. Electrons injected into the electron injection layer 28 are transported by the electron transport layer 27 to the emissive layer 26. Holes and electrons then recombine in the emissive layer 26, exciting the emissive material of the emissive layer 26 and causing the emissive layer 26 to emit light.

The light-emitting layers 26 are formed, for example, as a red light-emitting layer, a green light-emitting layer, and a blue light-emitting layer. The red light-emitting layer is formed from a red light-emitting material that emits red light, the green light-emitting layer is formed from a green light-emitting material that emits green light, and the blue light-emitting layer is formed from a blue light-emitting material that emits blue light. The red light-emitting layer, the green light-emitting layer, and the blue light-emitting layer are formed in the openings 31 of the bank 30.

The bank 30 separates the red, green, and blue light-emitting layer material liquids, preventing these material liquids from mixing. The bank 30 is insulating and fills the contact holes that penetrate the planarization layer 18.

Figure 2 is a plan view showing a substrate processing system according to one embodiment. The substrate processing system 100 forms a hole injection layer 24, a hole transport layer 25, and a light-emitting layer 26 on an anode 21. The substrate processing system 100 includes an input station 110, a processing station 120, an output station 130, and a control device 140.

The loading station 110 loads a cassette C containing multiple substrates 10 from the outside and sequentially removes the multiple substrates 10 from the cassette C. Each substrate 10 is pre-formed with a TFT layer 12, a planarization layer 18, an anode 21, a bank 30, etc.

The loading station 110 includes a cassette mounting table 111 on which a cassette C is placed, a transport path 112 provided between the cassette mounting table 111 and the processing station 120, and a substrate transport body 113 provided on the transport path 112. The substrate transport body 113 transports substrates 10 between the cassette C placed on the cassette mounting table 111 and the processing station 120.

The processing station 120 forms a hole injection layer 24, a hole transport layer 25, and an emissive layer 26 on the anode 21. The processing station 120 includes a hole injection layer formation block 121 that forms the hole injection layer 24, a hole transport layer formation block 122 that forms the hole transport layer 25, and an emissive layer formation block 123 that forms the emissive layer 26.

The hole injection layer formation block 121 forms the hole injection layer 24 by applying a material liquid for the hole injection layer 24 onto the anode 21 to form a coating layer, and then drying and baking the coating layer. The material liquid for the hole injection layer 24 contains an organic material and a solvent. The organic material may be either a polymer or a monomer. In the case of a monomer, it may be polymerized by baking to form a polymer.

The hole injection layer formation block 121 includes a coating device 121a, a buffer device 121b, a reduced-pressure drying device 121c, a heat treatment device 121d, and a temperature adjustment device 121e. The coating device 121a ejects droplets of the material liquid for the hole injection layer 24 toward the opening 31 of the bank 30. The buffer device 121b temporarily stores substrates 10 awaiting processing. The reduced-pressure drying device 121c dries the coating layer coated by the coating device 121a under reduced pressure to remove the solvent contained in the coating layer. The heat treatment device 121d heat-treats the coating layer dried by the reduced-pressure drying device 121c. The temperature adjustment device 121e adjusts the temperature of the substrate 10 heat-treated by the heat treatment device 121d to a predetermined temperature, for example, room temperature.

The coating device 121a, buffer device 121b, heat treatment device 121d, and temperature control device 121e maintain an atmospheric atmosphere inside. The reduced-pressure drying device 121c switches the internal atmosphere between atmospheric and reduced-pressure.

In the hole injection layer formation block 121, the arrangement, number, and internal atmosphere of the coating device 121a, buffer device 121b, reduced pressure drying device 121c, heat treatment device 121d, and temperature control device 121e can be selected as desired.

The hole injection layer formation block 121 also includes substrate transport devices CR1 to CR3 and transfer devices TR1 to TR3. The substrate transport devices CR1 to CR3 each transport the substrate 10 to an adjacent device. For example, the substrate transport device CR1 transports the substrate 10 to the adjacent coating device 121a and buffer device 121b. The substrate transport device CR2 transports the substrate 10 to the adjacent reduced pressure drying device 121c. The substrate transport device CR3 transports the substrate 10 to the adjacent heat treatment device 121d and temperature adjustment device 121e. The transfer devices TR1 to TR3 are respectively provided between the loading station 110 and the substrate transport device CR1, between the substrate transport device CR1 and the substrate transport device CR2, and between the substrate transport device CR2 and the substrate transport device CR3, and relay the substrate 10 between them. The interiors of the substrate transport devices CR1-CR3 and the transfer devices TR1-TR3 are maintained in an atmospheric environment.

A transfer device TR4 is provided between the substrate transfer device CR3 in the hole injection layer formation block 121 and the substrate transfer device CR4 in the hole transport layer formation block 122 to relay the substrate 10 between them. The interior of the transfer device TR4 is maintained at atmospheric pressure.

The hole transport layer forming block 122 forms the hole transport layer 25 by applying a material liquid for the hole transport layer 25 onto the hole injection layer 24 to form a coating layer, and then drying and baking the coating layer. The material liquid for the hole transport layer 25 contains an organic material and a solvent. The organic material may be either a polymer or a monomer. In the case of a monomer, it may be polymerized by baking to form a polymer.

The hole transport layer formation block 122 includes a coating device 122a, a buffer device 122b, a reduced-pressure drying device 122c, a heat treatment device 122d, and a temperature adjustment device 122e. The coating device 122a ejects droplets of the material liquid for the hole transport layer 25 toward the opening 31 of the bank 30. The buffer device 122b temporarily stores substrates 10 awaiting processing. The reduced-pressure drying device 122c dries the coating layer coated by the coating device 122a under reduced pressure to remove the solvent contained in the coating layer. The heat treatment device 122d heat-treats the coating layer dried by the reduced-pressure drying device 122c. The temperature adjustment device 122e adjusts the temperature of the substrate 10 heated by the heat treatment device 122d to a predetermined temperature, for example, room temperature.

The coating device 122a and buffer device 122b maintain an atmospheric atmosphere inside. Meanwhile, the heat treatment device 122d and temperature adjustment device 122e maintain a low-oxygen, low-dew-point atmosphere inside to prevent deterioration of the organic material in the hole transport layer 25. The reduced-pressure drying device 122c switches its internal atmosphere between a low-oxygen, low-dew-point atmosphere and a reduced-pressure atmosphere.

Here, a low-oxygen atmosphere refers to an atmosphere with an oxygen concentration lower than that of the atmosphere, for example, an atmosphere with an oxygen concentration of 10 ppm or less. Furthermore, a low-dew-point atmosphere refers to an atmosphere with a lower dew-point temperature than that of the atmosphere, for example, an atmosphere with a dew-point temperature of -10°C or less. A low-oxygen, low-dew-point atmosphere is formed, for example, with an inert gas such as nitrogen gas.

In the hole transport layer formation block 122, the arrangement, number, and internal atmosphere of the coating device 122a, buffer device 122b, reduced pressure drying device 122c, heat treatment device 122d, and temperature control device 122e can be selected as desired.

The hole transport layer formation block 122 also includes substrate transfer devices CR4-CR6 and transfer devices TR5-TR6. Substrate transfer devices CR4-CR6 transfer substrates 10 to their respective adjacent devices. Transfer devices TR5-TR6 are located between substrate transfer device CR4 and substrate transfer device CR5, and between substrate transfer device CR5 and substrate transfer device CR6, respectively, and relay substrates 10 between them.

The interior of substrate transport device CR4 is maintained in an atmospheric atmosphere. Meanwhile, the interiors of substrate transport devices CR5-CR6 are maintained in a low-oxygen, low-dew-point atmosphere. This is because the interior of reduced-pressure drying device 122c adjacent to substrate transport device CR5 can be switched between a low-oxygen, low-dew-point atmosphere and a reduced-pressure atmosphere. Furthermore, the interiors of heat treatment device 122d and temperature control device 122e installed adjacent to substrate transport device CR6 are maintained in a low-oxygen, low-dew-point atmosphere.

Transfer device TR5 is configured as a load lock device that switches its internal atmosphere between air and a low-oxygen, low-dew-point atmosphere. This is because a reduced-pressure drying device 122c is installed adjacent to transfer device TR6 downstream. Meanwhile, the interior of transfer device TR6 is maintained in a low-oxygen, low-dew-point atmosphere.

A transfer device TR7 is provided between the substrate transfer device CR6 in the hole transport layer formation block 122 and the substrate transfer device CR7 in the light emitting layer formation block 123 to relay the substrate 10 between them. The interior of the substrate transfer device CR6 is maintained in a low-oxygen, low-dew-point atmosphere, while the interior of the substrate transfer device CR7 is maintained in an atmospheric atmosphere. Therefore, the transfer device TR7 is configured as a load lock device that switches the atmosphere inside it between a low-oxygen, low-dew-point atmosphere and an atmospheric atmosphere.

The light-emitting layer forming block 123 forms the light-emitting layer 26 by applying a material liquid for the light-emitting layer 26 onto the hole transport layer 25 to form a coating layer, and then drying and baking the formed coating layer. The material liquid for the light-emitting layer 26 contains an organic material and a solvent. The organic material may be either a polymer or a monomer. In the case of a monomer, it may be polymerized by baking to form a polymer.

The light-emitting layer formation block 123 includes a coating device 123a, a buffer device 123b, a reduced-pressure drying device 123c, a heat treatment device 123d, and a temperature adjustment device 123e. The coating device 123a ejects droplets of the material liquid for the light-emitting layer 26 toward the opening 31 of the bank 30. The buffer device 123b temporarily stores substrates 10 awaiting processing. The reduced-pressure drying device 123c dries the coating layer coated by the coating device 123a under reduced pressure to remove any solvent contained in the coating layer. The heat treatment device 123d heat-treats the coating layer dried by the reduced-pressure drying device 123c. The temperature adjustment device 123e adjusts the temperature of the substrate 10 heated by the heat treatment device 123d to a predetermined temperature, for example, room temperature.

The interiors of the coating device 123a and buffer device 123b are maintained in an atmospheric atmosphere. On the other hand, the interiors of the heat treatment device 123d and temperature control device 123e are maintained in a low-oxygen, low-dew-point atmosphere to prevent deterioration of the organic material in the light-emitting layer 26. The reduced-pressure drying device 123c switches the interior atmosphere between a low-oxygen, low-dew-point atmosphere and a reduced-pressure atmosphere.

In the light-emitting layer formation block 123, the arrangement, number, and internal atmosphere of the coating device 123a, buffer device 123b, reduced-pressure drying device 123c, heat treatment device 123d, and temperature control device 123e can be selected as desired.

The light-emitting layer formation block 123 also includes substrate transport devices CR7-CR9 and transfer devices TR8-TR9. Substrate transport devices CR7-CR9 transport substrates 10 to their respective adjacent devices. Transfer devices TR8-TR9 are installed between substrate transport device CR7 and substrate transport device CR8, and between substrate transport device CR8 and substrate transport device CR9, respectively, to relay substrates 10 between them.

The interior of substrate transport device CR7 is maintained in an atmospheric atmosphere. Meanwhile, the interiors of substrate transport devices CR8-CR9 are maintained in a low-oxygen, low-dew-point atmosphere. This is because the interior of reduced-pressure drying device 123c adjacent to substrate transport device CR8 can be switched between a low-oxygen, low-dew-point atmosphere and a reduced-pressure atmosphere. Furthermore, the interiors of heat treatment device 123d and temperature control device 123e installed adjacent to substrate transport device CR9 are maintained in a low-oxygen, low-dew-point atmosphere.

Transfer device TR8 is configured as a load lock device that switches its internal atmosphere between air and a low-oxygen, low-dew-point atmosphere. This is because a reduced-pressure drying device 123c is installed adjacent to transfer device TR8 downstream. The interior of transfer device TR9 is maintained in a low-oxygen, low-dew-point atmosphere.

A transfer device TR10 is provided between the substrate transport device CR9 of the light-emitting layer formation block 123 and the unloading station 130 to relay the substrate 10 between them. The interior of the substrate transport device CR9 is maintained in a low-oxygen, low-dew-point atmosphere, while the interior of the unloading station 130 is maintained in an atmospheric atmosphere. Therefore, the transfer device TR7 is configured as a load lock device that switches the atmosphere inside it between a low-oxygen, low-dew-point atmosphere and an atmospheric atmosphere.

The unloading station 130 sequentially stores multiple substrates 10 in a cassette C and unloads the cassette C to the outside. The unloading station 130 includes a cassette mounting table 131 on which the cassette C is placed, a transport path 132 provided between the cassette mounting table 131 and the processing station 120, and a substrate transport body 133 provided on the transport path 132. The substrate transport body 133 transports substrates 10 between the processing station 120 and the cassette C placed on the cassette mounting table 131.

Next, a substrate processing method using the substrate processing system 100 configured as described above will be described. When a cassette C containing multiple substrates 10 is placed on the cassette mounting table 111, the substrate transport body 113 sequentially removes the substrates 10 from the cassette C on the cassette mounting table 111 and transports them to the hole injection layer formation block 121.

The hole injection layer formation block 121 applies the material liquid for the hole injection layer 24 onto the anode 21 to form a coating layer, and then dries and bakes the formed coating layer to form the hole injection layer 24. The substrate 10 on which the hole injection layer 24 has been formed is transferred from the hole injection layer formation block 121 to the hole transport layer formation block 122 by the transfer device TR4.

The hole transport layer forming block 122 applies a material liquid for the hole transport layer 25 onto the hole injection layer 24 to form a coating layer, and then dries and bakes the formed coating layer to form the hole transport layer 25. The substrate 10 on which the hole transport layer 25 has been formed is transferred from the hole transport layer forming block 122 to the light-emitting layer forming block 123 by the transfer device TR7.

The light-emitting layer forming block 123 applies the material liquid for the light-emitting layer 26 onto the hole transport layer 25 to form a coating layer, and then dries and bakes the formed coating layer to form the light-emitting layer 26. The substrate 10 on which the light-emitting layer 26 has been formed is transferred from the light-emitting layer forming block 123 to the carry-out station 130 by the transfer device TR10.

The substrate transport body 133 of the unloading station 130 places the substrate 10 received from the transfer device TR10 into a specified cassette C on the cassette mounting table 131. This completes the series of substrate 10 processing steps in the substrate processing system 100.

The substrate 10 is transported from the transport station 130 while still in the cassette C. An electron transport layer 27, an electron injection layer 28, a cathode 22, etc. are formed on the substrate 10 that has been transported to the outside.

[Control device configuration]
As shown in FIG. 3, the control device 140 includes an HDD 51, a ROM 52, a RAM 53, an I/O port 54, a CPU 55, a communication unit 56, and a bus 57 interconnecting these components.

The HDD 51 stores the OS and programs executed by the CPU 55. The HDD 51 can be replaced with an SSD or the like.

ROM 52 is a recording medium composed of EEPROM, flash memory, etc., that stores programs for starting up the control device.

RAM 53 functions as a work area for CPU 55. An operation panel 58, for example, is connected to I/O port 54. The operation panel 58 is a so-called touch panel display that displays the operation menu and the status of the droplet ejection device (remaining droplet amount, nozzle status, executed jobs, etc.). A keyboard may also be connected to the operation panel 58.

The CPU 55 controls the entire control device 140. The CPU 55 executes programs stored in the HDD 51 and sends signals to the droplet ejection device in accordance with instructions received by the operation panel 58, starting and stopping the ejection of droplets.

The communication unit 56 is connected to the LAN 59 and transmits information between the server, host computer, etc.

[Coating device and coating method]
Next, the coating device 123a of the light-emitting layer formation block 123 will be described mainly with reference to FIGS. 4 to 6. FIG. 4 is a plan view showing a coating device according to one embodiment. In FIG. 4, the loading position of the substrate 10 is indicated by a dashed line. FIG. 5 is a side view showing the state when the substrate is yawing by the substrate moving unit on the left side. In addition, in FIGS. 4 and 5, one side of the droplet discharge device 160 in the X direction is also referred to as the "left side," and the opposite side of the droplet discharge device 160 in the X direction is also referred to as the "right side."

The coating device 123a moves the landing position of droplets of functional liquid (e.g., material liquid for the light-emitting layer 26) on the substrate 10 in the X and Y directions to draw a functional liquid pattern on the substrate 10.

The coating device 123a includes, for example, a stage unit 150 that floats the substrate 10 at a predetermined height using gas pressure, and a droplet ejection device 160 that dispenses droplets of functional liquid onto the substrate 10 that is floating at the predetermined height from the stage unit 150. The coating device 123a also includes a substrate movement unit 170 that holds from above an edge of the substrate 10 (e.g., a portion outside the functional liquid drawing area) that is floating at the predetermined height from the stage unit 150 and moves it in the X and Y directions. The substrate movement units 170 are provided on both sides of the droplet ejection device 160 in the X direction. The multiple substrate movement units 170 hand off the substrate 10 and pass it under the droplet ejection device 160, while the droplet ejection device 160 dispenses droplets onto the substrate 10. The coating device 123a also includes a slide mechanism unit 200 that moves the droplet discharge device 160 in the Y direction, and a maintenance unit 210 that performs processing to maintain the functionality of the droplet discharge device 160.

The stage 150 has multiple gas inlets 151 on its top surface that spray gas. The stage 150 is connected to a gas supply source 152 that supplies gas to each gas inlet 151. When the gas supply source 152 is activated, gas is sprayed from each gas inlet 151 on the stage 150, and the gas pressure supports the substrate 10 at a constant height above the top surface of the stage 150.

The stage 150 may have multiple intake ports 153 on its top surface for sucking in gas. The stage 150 is connected to a gas suction source 154 that sucks in gas through each intake port 153. When the gas suction source 154 is activated, gas is sucked in through each intake port 153. This makes it possible to balance the amount of gas sucked in and the amount of gas injected, reducing variation in the gap between the substrate 10 and the stage 150 and improving the levelness of the top surface of the substrate 10.

As shown in FIG. 5, the stage section 150 may be divided into three regions X1, X2, and X3 in the X direction, with only the air supply port 151 provided in the regions X1 and X3 at both ends in the X direction, and both the air supply port 151 and the air intake port 153 provided in the region X2 in the center in the X direction. This can improve the levelness of the substrate 10 above the region X2 in the center in the X direction, and improve the accuracy of the functional liquid drawing pattern on the substrate 10. Note that both the air supply port 151 and the air intake port 153 may also be provided in the regions X1 and X3 at both ends in the X direction.

The stage unit 150 may have lift pins (not shown) that extend and retract from its upper surface. When the stage unit 150 transfers the substrate 10 between it and the robot, the lift pins protrude from the upper surface of the stage unit 150. On the other hand, when the substrate moving unit 170 moves the substrate 10 and the droplet ejection device 160 dispenses droplets of functional liquid onto the substrate 10, the lift pins retract into the upper surface of the stage unit 150. An air cylinder, for example, is used as the lifting mechanism for raising and lowering the lift pins.

The droplet ejection device 160 ejects droplets of functional liquid toward the substrate 10, which is suspended at a predetermined height from the stage unit 150. Multiple droplet ejection devices 160 (e.g., 10 in Figure 4) are lined up in the Y direction. The multiple droplet ejection devices 160 may be moved independently in the Y direction, or may be moved together in the Y direction.

Each droplet ejection device 160 has multiple ejection heads 161 (see Figure 5). Each ejection head 161 has, on its underside, an ejection nozzle row consisting of multiple ejection nozzles lined up in the Y direction. Each ejection head 161 may also have multiple ejection nozzle rows on its underside.

Each ejection head 161 has a piezoelectric element for each ejection nozzle. When a voltage is applied to the piezoelectric element, the piezoelectric element deforms and droplets are ejected from the ejection nozzle. A heater or the like may be used instead of the piezoelectric element. When a voltage is applied to the heater, bubbles are generated, and the pressure of the generated bubbles causes droplets to be ejected from the ejection nozzle.

Each droplet ejection device 160 may eject multiple types of functional liquid. Examples of multiple types of functional liquid include material liquid for a red light-emitting layer, material liquid for a green light-emitting layer, and material liquid for a blue light-emitting layer. Multiple ejection nozzles provided on the same ejection head 161 eject droplets of the same type of functional liquid.

Figure 6 is a plan view showing the arrangement of ejection heads according to one embodiment. Each droplet ejection device 160 has two ejection head rows 162 aligned in the Y direction. Each ejection head row 162 is composed of six ejection heads 161 aligned in a stepped pattern in the X direction. Each ejection head row 162 has two ejection heads 161R that eject droplets of material liquid for the red light-emitting layer, two ejection heads 161G that eject droplets of material liquid for the green light-emitting layer, and two ejection heads 161B that eject droplets of material liquid for the blue light-emitting layer.

As shown in FIG. 4, the slide mechanism unit 200 moves the droplet ejection device 160 between a position where it ejects droplets of functional liquid onto the substrate 10, which is suspended at a predetermined height above the stage unit 150, and a position where it accepts processing by the maintenance unit 210 to maintain its functionality. The slide mechanism unit 200 has a pair of Y-axis beams 201 suspended above the stage unit 150, a pair of Y-axis guides 202 mounted on the pair of Y-axis beams 201, and a pair of Y-axis linear motors that move the droplet ejection device 160 along the pair of Y-axis guides 202.

The maintenance unit 210 performs processes to maintain the functionality of the droplet ejection device 160 and resolves ejection defects in the droplet ejection device 160. The maintenance unit 210 has a wiping unit 211 that wipes the area around the ejection openings of the ejection nozzles, and a suction unit 212 that sucks droplets from the ejection openings of the ejection nozzles. The suction unit 212 also serves to block the ejection openings of ejection nozzles that are in an inactive state, preventing clogging due to drying.

[Impact position detection method]
In the present disclosure, one method for calculating the degree of nozzle abnormality is to detect the landing position of each nozzle. When measuring the landing position, a test film different from the substrate 10 to be produced is fed onto the stage unit 150. The test film may be changed depending on the type of droplet. The landing position is measured from image data obtained by capturing an image of this test film with a camera.

The expected deviation in landing position is smaller by an order of magnitude or more compared to the nozzle arrangement (distance between nozzles). For this reason, when performing a nozzle check, the droplet ejection device 160 ejects ink from all nozzles simultaneously, and an image of the test film is taken, for example, with a camera installed on the stage unit 150. The control device 140 (or any information processing device) uses image processing to detect position information (deviation from the reference position) of each droplet (landing) drawn on the test film corresponding to each nozzle from the image data. The reference position is the target ejection position.

[Method for capturing droplet flight images]
Furthermore, in the present disclosure, one method for calculating the degree of nozzle abnormality involves capturing images of droplets flying from each nozzle. FIG. 7 illustrates an example of a method for capturing droplet flight images. As illustrated in FIG. 7 , an imaging device 301 is installed so that its optical axis is perpendicular to the direction in which the ejection head 161 ejects droplets. In FIG. 7 , three ejection heads 161 are positioned in the depth direction relative to the optical axis of the imaging device 301, resulting in a total of six nozzle rows. The nozzle rows extend in a direction perpendicular to the plane of the page. When viewed from the imaging device 301, the six nozzles overlap in the depth direction. However, when the focal position is adjusted to each nozzle, droplets from other nozzles become blurred and invisible. The imaging device 301 can capture droplet flight images by capturing still images triggered by an ejection signal while changing the focal position. The imaging device 301 may also capture video. In the case of video, for example, a frame in which the tip of the droplet is positioned near the center of the image is extracted.

[Cleaning-related functions]
8 is a functional block diagram illustrating cleaning-related functions of the control device 140. The control device 140 has an acquisition unit 60 (landing position acquisition unit 61, droplet flight image acquisition unit 62), a nozzle check determination unit 63, a landing position time-series data calculation unit 64, an image time-series data calculation unit 65, a calculation unit 66, a determination unit 67, a display unit 68, an abnormality mode determination unit 69, a cleaning type determination unit 70, and a cleaning execution unit 71. These functions of the control device 140 are functions or means obtained when the CPU of the control device 140 executes a program.

The acquisition unit 60 acquires information related to the droplet ejection state. The acquisition unit 60 has a landing position acquisition unit 61 and a droplet flight image acquisition unit 62. The landing position acquisition unit 61 acquires the landing positions detected using the method described above. The landing positions are obtained in the order in which the droplet ejection device 160 ejects droplets, making it possible to calculate time-series data. The landing position acquisition unit 61 may acquire the landing position for each droplet ejection, or may acquire the landing position every certain number of droplets.

The droplet flight image acquisition unit 62 acquires droplet flight images captured using the method described above. The droplet flight images are acquired in the order in which the droplet ejection device 160 ejects droplets, making it possible to calculate time-series data. The droplet flight image acquisition unit 62 may acquire a droplet flight image for each droplet ejection, or may acquire droplet flight images every certain number of droplets.

The nozzle check determination unit 63 calculates the degree of abnormality, etc., and determines whether it is time to perform cleaning or schedule cleaning. Nozzle checks may be performed, for example, after processing a certain number of substrates 10, at certain intervals, at a fixed time, or at the user's command.

The time series data calculation unit 64 calculates time series data of the impact position, and the time series data calculation unit 65 calculates time series data of the feature amounts obtained from the flight image. These will be described in detail later.

The calculation unit 66 calculates the degree of abnormality for each nozzle based primarily on the time-series data. If a correlation is found in the time-series data, the calculation unit 66 calculates the degree of abnormality according to the degree of correlation, or calculates the degree of abnormality according to the amount of change in the time-series data. The degree of abnormality can be calculated before an abnormality occurs, making it possible to detect signs of an abnormality. The greater the degree of abnormality, the higher the value of the degree of abnormality.

The judgment unit 67 determines whether the degree of abnormality is equal to or greater than a threshold value, and determines whether there are signs of an abnormality or whether an abnormality has already occurred.

The display unit 68 displays the abnormality level of each nozzle on the operation panel 58 or the like in a manner corresponding to the abnormality level (in a manner corresponding to the parameters). The display unit 68 may send the abnormality level to a person in charge via email or the like. The display unit 68 also displays the number of nozzles with an abnormality level above a certain level.

The abnormal mode determination unit 69 uses time-series data to determine the abnormal mode. Abnormal modes mainly include one in which the feature values of the impact position or droplet flight image gradually deviate from normal values, and one in which the feature values suddenly deviate from normal values. In addition to these two, there are also abnormal modes that depend on the abnormality detection method.

The cleaning type determination unit 70 determines the cleaning type depending on the abnormality mode, or depending on the abnormality mode and the degree of abnormality. As will be described in detail later, for example, the cleaning intensity is determined depending on the abnormality mode.

The cleaning execution unit 71 performs cleaning of the type determined by the degree of abnormality and abnormality mode. If the degree of abnormality is low or in the precursory stage, cleaning is performed at a convenient time based on the manufacturing process of the substrate 10. Furthermore, if the degree of abnormality is high (if an abnormality has already occurred), the cleaning execution unit 71 performs cleaning early.

Cleaning may be performed on a nozzle-by-nozzle basis, or on a discharge head 161-by-discharge head basis. Cleaning may also be performed when an abnormality occurs in N or more nozzles (N is a natural number) of the discharge head 161.

[Time series data of impact position and determination of abnormality level]
Next, we will explain the time-series data of the landing positions and the method for determining the degree of abnormality with reference to FIGS. 9 to 11. FIG. 9 shows an example of time-series data obtained from the landing positions of droplets ejected by a normal nozzle. FIG. 10 shows an example of time-series data obtained from the landing positions of an abnormal nozzle. FIGS. 9(a) and 10(a) are graphs showing the temporal changes in the landing positions. The horizontal axis represents the number of data points. The vertical axis represents the x- and y-coordinates of the landing position, and the distance r from the origin. The origin is the target ejection position, which is predetermined for the test film. In FIG. 9(a), the landing positions vary around a constant value, but in FIG. 10(a), it can be seen that the landing positions gradually move in the negative direction.

Figures 9(b) and 10(b) show mapping data of impact positions. Mapping data shows impact positions on a two-dimensional plane in the x and y directions. The horizontal axis is the x-coordinate of the impact position, and the vertical axis is the y-coordinate of the impact position. Figures 9(b) and 10(b) are in black and white for ease of drawing, but the data points are displayed in different colors according to time period. In the mapping data of Figure 9(b), the impact positions are uniformly distributed, but in Figure 10(b), the impact positions tend to move in the lower left direction. Arrow 410 in Figure 10(b) indicates the direction of movement of the impact position, which changes over time.

If a human were to determine the tendency of droplet landing positions, a certain amount of data on droplet landing positions, as shown in Figures 10(a) and 10(b), would be required. This means that it would take a relatively long time to determine whether cleaning should be performed. Furthermore, the timing of the decision to perform cleaning can vary from person to person, and even within the same person.

Figures 9(c) and 10(c) are correlograms of impact positions. A correlogram is a graph of the correlation function (autocorrelation) of values at two points in time with a time difference of n, calculated as n increases. For ease of illustration, the graph is in black and white, but the correlation functions of the x and y coordinates are plotted against n. The horizontal axis represents n, and the vertical axis represents the correlation function (calculated value). A correlation function of 1.0 indicates perfect correlation, 0 indicates complete no correlation, and -1 indicates complete anti-correlation. Correlation between impact positions can be interpreted as, for example, when the data at time t is small, the data at time t+1 also tends to be small, or when the data at time t is large, the data at time t+1 also tends to be large. In other words, the correlogram can detect a trend toward movement in the impact position. In Figure 9(c), the correlation function is small regardless of n. On the other hand, Figure 10(c) shows that the correlation function is large in regions where n is small. In other words, it is possible to detect a tendency for impact positions to move with a small amount of impact position data; in other words, it is possible to determine early on that there is a high possibility that the impact position will shift (as a sign of an abnormality).

The criterion for determining whether the correlation function is large or small may be an arbitrary threshold value, or it may be determined whether the correlation function is outside or inside the confidence interval 411.

A method for calculating the degree of abnormality using a correlogram will be described. As an example, the calculation unit 66 uses Equation (1) to calculate the number of data p outside the confidence interval among n correlation functions.
The sum of the correlation functions of the data outside the confidence interval, q
is weighted by coefficients a and b to calculate the degree of abnormality.
Abnormality degree = a x p + b x q ... (1)
The method for calculating formula (1) is merely an example, and the degree of abnormality may be calculated using either p or q, or the degree of abnormality may be calculated from the value of the correlation function by changing n without considering the confidence interval. Furthermore, in order to detect signs of abnormality, the calculation unit 66 may calculate the degree of abnormality using only n that is less than a certain value.

The greater the autocorrelation of time series data, the greater the degree of anomaly. However, the absolute value of the distance must also be taken into account when determining the degree of anomaly. For example, [0, 1, 2, 3, 4] has a large autocorrelation but a small absolute value of the distance, while [0, 1, 2, 100, 101] has a small autocorrelation but a large absolute value of the distance. Therefore, it is a good idea to take the absolute value of the distance into account in equation (1).

Next, we will explain how to calculate the degree of abnormality when the impact position suddenly changes. A correlogram can detect early signs of an abnormality when the impact position moves over time, but some abnormalities involve a sudden change in the impact position. We will explain how to calculate the degree of abnormality when such an abnormality occurs.

Figure 11 is a graph showing changes in impact position over time. The horizontal axis represents the number of data points, and the vertical axis represents the x- or y-coordinate of the impact position. As shown in Figure 11, it can be seen that the impact position changes suddenly at data points na and nb.

Such abnormalities can be detected by calculating the differences 310 in impact positions over time and comparing the differences 310 with a threshold value. The degree of abnormality may be the number of differences that are equal to or greater than the threshold value, the value obtained by dividing the number of differences that are equal to or greater than the threshold value by n, the sum of n differences 310, etc.

Mechanically learning the degree of anomaly is also effective. By inputting time-series data such as those shown in Figures 9(a), 10(a), and 11 into a neural network and learning using human annotations (the degree of anomaly determined by humans) as teaching signals, it is possible to construct a classification device that calculates the degree of anomaly for input time-series data. The learning method for machine learning may be any of supervised learning, unsupervised learning, semi-supervised learning, reinforcement learning, and deep learning, or may be a combination of these learning methods. Any learning method for machine learning may be used. Machine learning techniques include perceptrons, deep learning, support vector machines, logistic regression, naive Bayes, decision trees, and random forests, and are not limited to the methods described in this embodiment.

[Time series data of droplet flight images and degree of anomaly]
Next, time-series data of feature quantities acquired from droplet flight images will be described with reference to Figures 12 to 18. In this disclosure, feature quantities will be described as bending, droplet size, droplet spot, nozzle plate contamination, nozzle clog, and deviation in landing time for each nozzle. It is not necessary to use all of these feature quantities; it is sufficient to use one or more feature quantities.

Note that one droplet flight image shows multiple droplets. To distinguish these nozzles one by one, the time-series data calculation unit 65 may separate the droplet flight image at fixed horizontal distances. The horizontal coordinates of the droplet flight image and the nozzle identification information (which nozzle is shown in which range of the droplet flight image) are associated in advance. Therefore, the feature values of the droplet flight image are detected for each nozzle.

Bending Figure 12 shows droplet flight images of different droplets ejected from the same nozzle as the droplet flight image. As described above, the droplet flight images are images of droplets flying from the nozzle toward the substrate 10. Time has passed in the order of Figure 12(a) → Figure 12(b) → Figure 12(c).

The time series data calculation unit 65 binarizes the droplet flight image and detects a straight line 311 in the flight direction. To detect the line, for example, a Hough transform or the least squares method may be used. The time series data calculation unit 65 calculates time series data for the slope θ of this line 311.

The degree of abnormality based on bending can be determined from correlograms and differences between data, just like with time series data on impact positions.

Figure 13 shows an example of time-series data using curvature, a feature quantity detected from droplet flight images. Figure 13(a) shows the temporal progression of droplets from the same nozzle captured in the droplet flight image. The angle relative to the ejection direction increases over time.

Figure 13(b) is a graph showing the change 321 in the angle of droplets from the same nozzle over time. Figure 13(b) also shows a schematic representation of a correlogram 322. Note that the droplet flight image in Figure 13(a) corresponds to the time axis in Figure 13(b). There is almost no bending up to times t1 and t2, but the droplets begin to bend from time t3 onwards. Correlogram 322 is also constant up to time t2, but begins to increase from time t2 onwards. Therefore, by referring to correlogram 322, it becomes easier to detect abnormalities even when the angle of bending is small. In other words, it is possible to detect signs of abnormalities.

As shown in Figure 13, the flight direction of the droplets may curve in two or more straight lines. In this case, the Hough transform detects each straight line, so if two or more straight lines are detected, the judgment unit 67 may determine that an abnormality exists.

Droplet Size FIG. 14 shows images of droplet flight ejected from several nozzles. To acquire droplet size, the time-series data calculation unit 65 binarizes the droplet flight image and calculates the area of black pixels (the number of black pixels) for each droplet trajectory. The time-series data calculation unit 65 calculates time-series data for this area. As shown in FIG. 14, droplets in the droplet flight image surrounded by dotted line 331 are small. Therefore, the droplet size can be estimated from the area of the black pixels representing the droplet trajectory.

The degree of abnormality based on droplet size can be determined from correlograms and differences between data, as well as time series data on impact positions.

Droplet Spots Figure 15 shows a droplet flight image including droplet spots. When acquiring droplet spots, the time-series data calculation unit 65 calculates the difference between a normal droplet flight image stored in advance and the captured droplet flight image. In other words, the difference between pixels at the same pixel position in the normal droplet flight image and the captured droplet flight image is found. The time-series data calculation unit 65 calculates the time-series data of the sum of these differences.

Figure 15(a) shows a droplet flight image with a droplet spot 340, and Figure 15(b) shows a normal droplet flight image. Figure 15(c) shows the difference image between Figure 15(a) and Figure 15(b). The difference image makes the droplet spot 340 clear. The judgment unit 67 judges that a droplet spot 340 is present if there are black pixels with an area equal to or greater than a certain level.

Since droplet spot 340 indicates the presence of a droplet in a space where no droplets normally exist, when the determination unit 67 detects droplet spot 340, it may immediately determine that an abnormality exists. The degree of abnormality of droplet spot 340 is set to a predetermined level based on knowledge of whether the droplet spot can be easily restored.

Since droplet spots 340 may only be detected temporarily, if droplet spots 340 are detected a certain number of times or more within a certain period of time, the judgment unit 67 may determine that an abnormality has occurred.

- Nozzle plate contamination Figure 16 is a droplet flight image that includes contamination on the nozzle plate. The nozzle plate contamination 350 is droplets that have adhered to the vicinity of the nozzle. If the nozzle plate becomes contaminated, it may eventually cause nozzle clogging or other problems. As with droplet spots, the method for detecting nozzle plate contamination uses a difference image. The time-series data calculation unit 65 calculates a difference image between a normal droplet flight image stored in advance and the droplet flight image. In other words, it calculates the difference between pixels at the same pixel position. The time-series data calculation unit 65 calculates the total area of black pixels included in this difference image and calculates the total time-series data.

The degree of abnormality due to nozzle plate contamination can be determined from correlograms and differences between data, just like with time-series data on impact positions.

Missing nozzles Figure 17 is a droplet flight image that includes a missing nozzle. Missing nozzles 360 refer to a nozzle not ejecting droplets. Possible causes include a clogged nozzle. Nozzle contamination can be easily detected by checking whether or not there is a straight line representing the droplet trajectory. The time-series data calculation unit 65 binarizes the portion of one nozzle in the droplet flight image and determines whether or not it can be considered to have no black pixels.

Nozzle clogs mean that droplets that should be present are not present, so when the judgment unit 67 detects a nozzle clog, it can immediately conclude that an abnormality exists. The degree of abnormality of the nozzle clog is determined to be a preset abnormality level based on knowledge of whether the nozzle clog can be easily repaired.

Since nozzle missing may only be detected temporarily, the judgment unit 67 may judge the degree of abnormality if a certain number of nozzle missing events are detected within a certain period of time.

Differences in landing time between nozzles Figure 18(a) is a droplet flight image that explains the difference in landing time between nozzles. Figure 18(b) shows an enlarged view of the tip of a droplet. The time it takes for droplets from nozzles that are simultaneously discharge-controlled to reach the substrate 10 can vary depending on the nozzle. This is presumably due to poor discharge caused by nozzle clogging.

The difference in impact time for each nozzle can be detected from the droplet tip position 370, as shown in Figure 18 (b). The time-series data calculation unit 65 binarizes each nozzle portion of the droplet flight image and detects the coordinates of the tip position 370 for each nozzle. The time-series data calculation unit 65 calculates the difference in the coordinates of the tip position 370 for each combination of two nozzles and determines the maximum value of all the differences. The time-series data calculation unit 65 calculates the time-series data of this maximum difference.

The degree of abnormality based on the deviation in impact time for each nozzle can be determined from the correlogram and differences between data, just like the time series data for impact position.

[Determining abnormal mode]
Next, a method for determining an abnormal mode using time-series data will be described with reference to Fig. 19. Fig. 19 shows time-series data of landing positions, but the same applies to time-series data obtained from droplet flight images.

Figures 19(a) to (d) are graphs showing changes in impact position over time. Of these, Figures 19(a) and (b) represent an abnormal mode in which the impact position changes suddenly, while Figures 19(c) and (d) represent an abnormal mode in which the impact position changes gradually.

Figures 19(e) to (h) show mapping data for impact positions. Of these, Figures 19(e) and (f) show abnormal modes in which the impact position changes suddenly, while Figures 19(g) and (h) show abnormal modes in which the impact position changes gradually.

The abnormal mode determination unit 69 calculates the difference between two adjacent impact positions (t and t+1) from graphs showing changes in impact position over time, such as those shown in Figures 19(a) to 19(d). If the difference (absolute value) exceeds a threshold, the abnormal mode determination unit 69 determines that an abnormal mode has occurred, in which the impact position suddenly changes.

The abnormal mode determination unit 69 also clusters the mapping data shown in Figures 19(e) to (h) using the k-means method or similar. In the case of an abnormal mode in which the impact position changes suddenly, as shown in Figures 19(e) and (f), the impact positions can be separated into two clusters (areas).

Figure 20 is a diagram explaining how to determine whether classification into a cluster has been successful. As shown in Figure 20(a), it is assumed that the abnormality mode determination unit 69 has classified the data into cluster k=2.

Next, as shown in Figure 20(b), the abnormality mode determination unit 69 calculates a line 383 that passes through the centers of gravity 381, 382 of the two clusters. It also calculates a line 384 that passes through the midpoint of line 383 and is perpendicular to line 383. At points on line 384, g(x) in equation (2) becomes zero. Therefore, if the two clusters form separate groups, line 384 becomes a discriminant function for linear separation. In equation (2), _w0 to _w2 are coefficients of the line.
g(x)=w ₀ + w ₁ x + w ₂ y ……(2)
The abnormality mode determination unit 69 sets the actual measured values of the impact positions of cluster 1 and cluster 2 to x and y, respectively, and can determine that the impact positions have been separated (the impact positions have changed suddenly) if the proportion of positive g(x) due to the impact positions of cluster 1 is equal to or greater than a threshold value, and the proportion of negative g(x) due to the impact positions of cluster 2 is equal to or greater than a threshold value.

There are various clustering techniques available, including hierarchical techniques and EM algorithms.

Abnormal modes include a gradual change in impact position, a sudden change in impact position, droplet spots, missing nozzles, and deviations in impact time between nozzles. These are detected by abnormal modes. It is also possible to determine the degree of abnormality in the deviations in impact time between nozzles using time-series data.

[Determining cleaning type based on abnormality mode]
If the impact position suddenly changes, it is likely that debris (solidified droplets, etc.) has suddenly adhered to the nozzle. For this reason, the cleaning type determination unit 70 determines that the debris can be removed with light cleaning. It is known that there are cases where the nozzle can return to normal without cleaning.

- If the landing position changes gradually, it is assumed that dust is adhering continuously (persistent adhesion). The cleaning type determination unit 70 determines that the cleaning intensity needs to be increased.

In this way, by classifying abnormality modes, if it is determined that light cleaning is sufficient, the amount of (expensive) droplets used for cleaning can be reduced, thereby reducing maintenance costs.

This concept also applies when an abnormal mode is determined based on time-series data of the feature quantities of droplet flight images.

When droplet spots, missing nozzles, and differences in landing time between nozzles are detected, the cleaning type determination unit 70 determines the cleaning type with a strength that corresponds to the set abnormality level. If the difference in landing time between nozzles gradually progresses, it is believed that nozzles with slow landing times are experiencing ejection problems due to liquid clogging. It is preferable that the cleaning type determination unit 70 change the cleaning strength depending on the size of the time lag.

It is also preferable that the cleaning type determination unit 70 determine the cleaning type taking into consideration not only the abnormal mode but also the degree of abnormality. For example, the cleaning type determination unit 70 increases or decreases the number of cleaning runs and duration for the cleaning type determined in the abnormal mode depending on the degree of abnormality.

[Operation procedure]
21 is a flowchart showing the procedure for the control device 140 to monitor nozzle abnormalities. The nozzle check determination unit 63 determines whether it is time to perform a nozzle check (S1). The timing for a nozzle check may be, for example, when discharge for a certain number of substrates 10 has finished, or when a certain amount of time has passed.

The test film may be placed between a certain number of substrates 10, or dispensing onto the substrates 10 may be completed before dispensing onto the test film. The landing position acquisition unit 61 detects the landing position (x, y coordinates) from the image data of the test film. Since the test film is not required for droplet flight images, the imaging device captures droplet flight images as appropriate during dispensing onto the substrates 10. The droplet flight image acquisition unit 62 stores droplet flight images as appropriate. The droplet flight images may also be captured during dispensing onto the test film.

If it is time to perform a nozzle check, the time-series data calculation unit 64 calculates time-series data of the landing position (S2). Similarly, the time-series data calculation unit 65 calculates time-series data of the feature amounts of the droplet flight image. Time-series data does not need to be calculated for feature amounts such as droplet spots and nozzle misses.

Next, the calculation unit 66 calculates the degree of abnormality using equation (1) and a correlogram, which is a type of time-series data (S3). The degree of abnormality is calculated for each of the impact position and the droplet flight image. Furthermore, for the droplet flight image, it can be calculated for each feature amount (curvature, droplet size, contamination on the nozzle plate, and deviation in impact time for each nozzle). Therefore, the calculation unit 66 weights each degree of abnormality to calculate the final degree of abnormality. Alternatively, it may calculate each degree of abnormality in advance and determine whether any one of the degrees of abnormality exceeds a threshold.

The judgment unit 67 determines whether the degree of abnormality is equal to or greater than the threshold value (S4). If the answer is No in step S4 (if the degree of abnormality is not equal to or greater than the threshold value), the nozzle check is performed again from step S1.

It is also preferable that this threshold be set in two stages: a first threshold corresponding to a sign of an abnormality, and a second threshold corresponding to an abnormality that has already occurred and requires immediate maintenance. If the value is equal to or greater than the first threshold (less than the second threshold), the cleaning execution unit 71 can schedule and perform maintenance. If the value is equal to or greater than the second threshold, the cleaning execution unit 71 can perform maintenance immediately.

If step S4 returns Yes, the abnormality mode determination unit 69 determines the abnormality mode (S5). That is, it determines whether the impact position, droplet curvature, droplet size change, nozzle plate contamination, or deviation in impact time between nozzles is occurring gradually or suddenly. Furthermore, it determines whether droplet spots or missing nozzles are occurring based on the feature amounts of the droplet flight image.

Then, the cleaning type determination unit 70 determines the cleaning type according to the abnormality mode and the degree of abnormality (S6). For example, if the abnormality mode occurs suddenly, it may determine that light cleaning should be performed, and if the abnormality mode occurs gradually, it may determine that strong cleaning should be performed. Furthermore, it may be considered that the number of cleanings or the time period may be increased according to the degree of abnormality. The cleaning type determination unit 70
The type of cleaning can be determined by combining one or more of the cleaning ejection intensity, the number of cleaning ejections, and the cleaning ejection time.

The cleaning type determination unit 70 may also prepare a table that associates cleaning types with abnormality levels and abnormality modes, and use this to determine the cleaning type.

The cleaning execution unit 71 performs the determined cleaning type based on the maintenance schedule (S7).

The control device 140 determines whether to stop operation of the droplet ejection device 160 (S8). This determination is made, for example, based on a nozzle check after cleaning is performed.

As shown in Figure 22, the timing of the nozzle check may be determined and similar processing may be performed. Figure 22 is a flowchart showing the procedure by which the control device 140 monitors nozzle abnormalities. The explanation of Figure 22 will mainly focus on the differences from Figure 21.

In the process of Figure 22, the control device 140 stores the landing positions and droplet flight images in advance, so in step S11 the control device 140 deletes the old landing positions and droplet flight images (S11).

Next, the landing position acquisition unit 61 detects and saves the landing position (x, y coordinates) from the image data of the test film, and the droplet flight image acquisition unit 62 saves the droplet flight image (S12).

The calculation unit 66 determines whether the number of times the impact positions and droplet flight images have been saved has reached a specified number (S13). If the determination in step S13 is No, the process returns to step S12, and the impact positions and droplet flight images are repeatedly saved until the specified number of times has been reached.

If the determination in step S13 is Yes, the control device 140 executes steps S14 to S20. These processes may be similar to steps S2 to S8 in Figure 21. However, if the determination in step S16 is No, the process returns to step S12.

[Display of nozzles where abnormalities were detected]
23 is an example of a nozzle monitoring screen 390 displayed on the operation panel 58 connected to the control device 140. The nozzle monitoring screen 390 has a discharge head 161 for each nozzle row, and shows the status of the nozzles in each discharge head 161. The number of nozzles, number of rows, number of heads, etc., in one discharge head 161 are merely examples.

On the nozzle monitoring screen 390, nozzles 391-393 with an abnormality level above a threshold (including cases where there are signs of an abnormality) are highlighted. For example, they may be displayed in a different color than normal nozzles or may blink. This allows the user to easily determine which nozzle is abnormal. When the user presses a highlighted nozzle 391-393, a pop-up will appear, showing the abnormality level and abnormality mode, for example.

When nozzle cleaning is performed for each ejection head 161, the controller 140 can determine whether or not to perform cleaning for each ejection head 161, since the malfunctioning nozzles and their locations are known for each ejection head 161.

The nozzle monitoring screen 390 also displays the number of nozzles 394 whose abnormality level is above the threshold, and the number of nozzles 395 that show signs of abnormality. The user can check the number of nozzles whose abnormality level is above the threshold and stop manufacturing the substrate 10 or perform cleaning. The user can also check the number of nozzles that show signs of abnormality and plan cleaning schedules.

[Major Effects]
As described above, the droplet ejection device disclosed herein can optimize cleaning frequency by determining the timing of cleaning based on the degree of abnormality. Furthermore, by optimizing cleaning frequency and performing cleaning according to the abnormality mode, the amount of material consumed during cleaning can be reduced. For example, materials used for substrates of liquid crystal displays using organic electroluminescence (EL) are expensive, but the amount of material consumed during cleaning can be minimized. Furthermore, if signs of abnormality can be detected, it becomes easier to schedule processes for the droplet ejection device, and cleaning can be performed before an abnormality occurs, reducing the need to redo the formation of the target object.

[Other application examples]
The best mode for carrying out the present disclosure has been described above using examples, but the present disclosure is not limited to these examples in any way, and various modifications and substitutions can be made within the scope that does not deviate from the gist of the present disclosure.

For example, the function of detecting signs of abnormality may be separate from the droplet ejection device 160. The abnormality monitoring device that detects signs of abnormality may be connected to the droplet ejection device 160 via a USB cable or a network, for example. Alternatively, the abnormality monitoring device may be a server device that communicates with the droplet ejection device 160 via a network.

While the present embodiment describes an example in which the droplet ejection device 160 forms electrodes of an organic EL display, the droplet ejection device 160 may simply eject droplets. The ejected liquid is not particularly limited as long as it has a viscosity and surface tension that allows it to be ejected from the ejection head 161. However, it is preferable that the viscosity of the ejected liquid be reduced to a certain level at room temperature and pressure, or by heating or cooling. More specifically, the liquid may be a solution, suspension, emulsion, or the like containing a solvent such as water or an organic solvent; a colorant such as a dye or pigment; a functionalizing material such as a polymerizable compound, resin, or surfactant; a biocompatible material such as DNA, amino acids, proteins, or calcium; or an edible material such as a natural dye. These liquids can be used, for example, as inkjet inks, surface treatment liquids, liquids for forming components of electronic elements or light-emitting elements, or resist patterns for electronic circuits, and material liquids for three-dimensional modeling.

Furthermore, the configuration examples in Figure 8 and elsewhere are divided according to main functions to make it easier to understand the processing performed by the droplet ejection device 160. The method of dividing the processing units or their names does not limit the scope of this disclosure. The processing performed by the droplet ejection device 160 can also be divided into even more processing units depending on the processing content. Furthermore, it can also be divided so that one processing unit includes even more processing.

REFERENCE SIGNS LIST 10 Substrate 140 Control device 150 Stage unit 61 Landing position acquisition unit 62 Droplet flight image acquisition unit 64, 65 Time series data calculation unit 66 Abnormality degree calculation unit

Claims (13)

A droplet ejection device that ejects droplets from a nozzle,
an acquisition unit that acquires information about the droplet ejection state;
a time-series data calculation unit that calculates time-series data of information about the ejection state of the droplets;
a calculation unit that calculates a parameter related to the nozzle cleaning based on the time-series data;
a determination unit that determines the timing of cleaning the nozzle based on the parameter ,
the information about the ejection state of the droplets is a droplet flight image;
The droplet ejection device is characterized in that the calculation unit calculates feature amounts detected from the droplet flight image, and calculates the parameters based on time-series data of the feature amounts . an abnormality mode determination unit that determines an abnormality mode of the nozzle based on the time series data;
a cleaning type determination unit that determines a cleaning type for the nozzle according to the abnormality mode;
2. The droplet ejection device according to claim 1, further comprising: the abnormality mode determination unit determines an abnormality mode in which an abnormality in the nozzle suddenly occurs and an abnormality mode in which an abnormality in the nozzle gradually occurs,
3. The droplet ejection device according to claim 2, wherein the type of cleaning is determined in accordance with the abnormality mode. Representing the time series data as a correlogram;
4. The droplet ejection device according to claim 1, wherein the calculation unit calculates the parameter in accordance with the magnitude of autocorrelation of time-series data. The information about the droplet ejection state is the droplet landing position,
5. The droplet ejection device according to claim 1, wherein the calculation unit calculates the parameters based on time-series data of landing positions of the droplets. the information about the ejection state of the droplets is a droplet flight image;
The droplet ejection device according to any one of claims 1 to 4, wherein the calculation unit calculates a correlogram of feature amounts detected from the droplet flight image, and calculates the parameters based on the correlogram. the characteristic amount is a flight angle of the droplets, a droplet size, contamination of the nozzle plate, or a difference in landing time for each nozzle;
The droplet ejection device according to claim 6, wherein the calculation unit calculates the parameters based on a correlogram calculated from the angle at which the droplets fly, the droplet size, contamination on the nozzle plate, or the difference in landing time between nozzles. a display unit that displays a screen on which the nozzles are schematically arranged,
8. The droplet ejection device according to claim 1, wherein the display unit displays the nozzle in a manner corresponding to the parameter. The droplet ejection device described in any one of claims 4 to 7, characterized in that the judgment unit compares the parameter with a threshold value to detect signs of abnormality in the nozzle. The droplet ejection device according to claim 3 , wherein the cleaning type determination unit determines the cleaning type based on one of a cleaning ejection intensity, a cleaning ejection count, and a cleaning ejection time, or a combination of two or more of these. 4. The droplet ejection device according to claim 3, wherein the cleaning type determining unit determines the cleaning type based on a table in which the cleaning type is associated with the parameter and the abnormality mode. A parameter calculation method performed by a droplet ejection device that ejects droplets from a nozzle, comprising:
acquiring information about the droplet ejection state;
calculating a parameter related to nozzle cleaning based on the information about the droplet ejection state;
and determining the timing of cleaning based on the parameter,
the information about the ejection state of the droplets is a droplet flight image;
The calculating step is a parameter calculation method in which feature amounts detected from the droplet flight image are calculated, and the parameters are calculated based on time-series data of the feature amounts . A substrate processing system having the droplet ejection device described in any one of claims 1 to 9.