Diagnostic device, mounting device, and semiconductor device manufacturing method

Abstract

To provide a technique capable of obtaining data used for machine learning for performing damage prediction diagnosis.SOLUTION: A diagnostic device includes a data collection device that acquires a sensor signal from a sensor at the time of diagnosis for each axis and collects first waveform data, and a computer configured to generate second waveform data by processing the collected first waveform data, accumulate the second waveform data, and perform breakage diagnosis of diagnosing presence or absence of breakage by comparing normal waveform data for each device body measured before production with the second waveform data.SELECTED DRAWING: Figure 10

Description

[Technical Field]

[0001] The present disclosure relates to a diagnostic device and is applicable to, for example, a diagnostic device for a die bonder. [Background technology]

[0002] In one step of the manufacturing process of a semiconductor device, a die is picked up by a bond head in a die bonder and bonded to a substrate. It has been proposed to detect abnormalities by detecting vibrations of the bond head during this bonding operation (for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2019-50295 A Summary of the Invention [Problem to be solved by the invention]

[0004] An object of the present disclosure is to provide a technology that can obtain data to be used for machine learning to perform predictive damage diagnosis.

[0005] Other objects and novel features will become apparent from the description of this specification and the accompanying drawings. [Means for solving the problem]

[0006] A brief summary of representative aspects of this disclosure is as follows. That is, the diagnostic device includes a data collection device that acquires sensor signals from sensors at the time of diagnosis for each axis and collects first waveform data, and a computer that is configured to process the collected first waveform data to generate second waveform data, store the second waveform data, and perform a damage diagnosis to diagnose the presence or absence of damage by comparing the second waveform data with normal waveform data measured for each individual device before production. [Effects of the Invention]

[0007] According to the present disclosure, it is possible to obtain data to be used for machine learning to perform predictive damage diagnosis. [Brief explanation of the drawings]

[0008] [Figure 1] FIG. 1 is a schematic top view showing an example of the configuration of a die bonder according to an embodiment. [Figure 2] FIG. 2 is a diagram illustrating the schematic configuration when viewed from the direction of arrow A in FIG. [Figure 3] FIG. 3 is a block diagram showing a schematic configuration of a control system of the die bonder shown in FIG. [Figure 4] FIG. 4 is a flowchart showing a method for manufacturing a semiconductor device using the die bonder shown in FIG. [Figure 5] FIG. 5 is a diagram showing the mounting positions of the sensors in the embodiment. [Figure 6] FIG. 6 is a diagram showing the detection directions of angular velocity and acceleration of the sensor shown in FIG. [Figure 7] FIG. 7 is a block diagram showing an example of the configuration of a system using a diagnostic device according to an embodiment. [Figure 8] FIG. 8 is a block diagram showing an example of the configuration of the PLC shown in FIG. [Figure 9] FIG. 9 is a flow chart for explaining the relationship between production and diagnosis in the die bonder shown in FIG. [Figure 10] FIG. 10 is a flowchart showing a method for diagnosing the die bonder shown in FIG. [Figure 11] FIG. 11 is a diagram showing an example of first waveform data acquired multiple times in the early stages. [Figure 12] FIG. 12 is a diagram showing FFT data obtained by FFT processing the first waveform data shown in FIG. [Figure 13] Fig. 13(a) is a diagram showing averaged data obtained by calculating the average value of the FFT data shown in Fig. 12. Fig. 13(b) is a diagram showing averaged data obtained by calculating the average value of data subjected to FFT processing at the time of damage. [Figure 14] FIG. 14 is a diagram showing second waveform data calculated based on the averaged data shown in FIGS. 13(a) and 13(b). [Figure 15] Fig. 15(a) is a diagram illustrating the threshold value for determining damage in the frequency band of 100-200 Hz, and Fig. 15(b) is a diagram illustrating the threshold value for determining damage in the frequency band of 700-800 Hz. DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, embodiments will be described with reference to the drawings. However, in the following description, the same components will be assigned the same reference numerals, and repeated description may be omitted. Note that, in order to clarify the description, the width, thickness, shape, etc. of each part may be shown schematically compared to the actual embodiment. Furthermore, the dimensional relationships, ratios, etc. of each element between multiple drawings do not necessarily match.

[0010] The configuration of a die bonder, which is one embodiment of a semiconductor manufacturing apparatus (mounting apparatus), will be described with reference to Figures 1 and 2. Figure 1 is a schematic top view showing an example of the configuration of the die bonder in the embodiment. Figure 2 is a diagram illustrating the schematic configuration as seen from the direction of arrow A in Figure 1.

[0011] The die bonder 1 broadly comprises a wafer supply unit 10, a pickup unit 20, an intermediate stage unit 30, a bonding unit 40, a transport unit 50, a substrate supply unit 60, a substrate unloading unit 70, and a control unit (control device) 80. The Y2-Y1 direction is the front-to-rear direction of the die bonder 1, the X2-X1 direction is the left-to-right direction, and the Z1-Z2 direction is the up-to-down direction. The wafer supply unit 10 is located on the front side of the die bonder 1, and the bonding unit 40 is located on the rear side.

[0012] The wafer supply unit 10 includes a wafer cassette lifter 11, a wafer holder 12, a peeling unit 13, and a wafer recognition camera 14.

[0013] A wafer cassette lifter 11 moves a wafer cassette (not shown), which stores multiple wafer rings WR, up and down to the wafer transport height. A wafer correction chute (not shown) aligns the wafer rings WR supplied from the wafer cassette lifter 11. A wafer extractor (not shown) removes wafer rings WR from the wafer cassette and supplies them to the wafer holder 12, or removes them from the wafer holder 12 and stores them in the wafer cassette.

[0014] A wafer W is adhered (attached) onto a dicing tape DT, and the wafer W is divided into multiple dies D. The dicing tape DT is held by a wafer ring WR. The wafer W is, for example, a semiconductor wafer, and the dies D are semiconductor chips. A film-like adhesive material DF called a die attach film (DAF) is adhered between the wafer W and the dicing tape DT. The adhesive material DF hardens when heated.

[0015] The wafer holder 12 is moved in the X1-X2 and Y1-Y2 directions by a drive unit (not shown), and moves the die D to be picked up to the position of the peeling unit 13. The wafer holder 12 also rotates the wafer ring WR in the XY plane by a drive unit (not shown). The peeling unit 13 moves in the vertical direction by a drive unit (not shown). The peeling unit 13 adsorbs the dicing tape DT and peels the die D from the dicing tape DT by, for example, pushing up a block (not shown).

[0016] The wafer recognition camera 14 captures an image of the die D to be picked up from the wafer W, recognizes the pick-up position of the die D, and inspects the surface of the die D.

[0017] The pickup unit 20 has a pickup head 21 and a pickup head table 23. The pickup head 21 is provided with a collet 22 that suction-holds the peeled die D at its tip. The pickup head 21 picks up the die D from the wafer supply unit 10 and places it on the intermediate stage 31. The pickup head table 23 moves the pickup head 21 in the Z1-Z2 direction, the Y1-Y2 direction, and the X1-X2 direction. The pickup head table 23 may also rotate the pickup head 21.

[0018] The intermediate stage unit 30 has an intermediate stage 31 on which the die D is placed, and a stage recognition camera 34 for recognizing the die D on the intermediate stage 31. The intermediate stage 31 has suction holes that adsorb the placed die D. The placed die D is temporarily held on the intermediate stage 31. The intermediate stage 31 is both a placement stage on which the die D is placed and a pickup stage on which the die D is picked up.

[0019] The bonding section 40 includes a bond head 41, a bond head table 43, a substrate recognition camera 44, and a bond stage 46. The bond head 41 is provided with a collet 42 that suction-holds a die D at its tip. The bond head table 43 moves the bond head 41 in the Z1-Z2, Y1-Y2, and X1-X2 directions. The bond head table 43 may also rotate the bond head 41. The substrate recognition camera 44 captures an image of the substrate S and recognizes the bond position. Here, the substrate S may be, for example, a wiring board or a lead frame. The substrate S has multiple product areas (hereinafter referred to as package areas P) that will eventually become a single package. The substrate S also has position recognition marks (not shown) for the package areas P. The bond stage 46 is raised when the die D is placed on the substrate S, supporting the substrate S from below. The bond stage 46 has a suction port (not shown) for vacuum-adsorbing the substrate S, and is capable of fixing the substrate S. The bond stage 46 also has a heating unit (not shown) for heating the substrate S.

[0020] With this configuration, the pick-up position and posture are corrected based on the image data of the stage recognition camera 34, and the bond head 41 picks up the die D from the intermediate stage 31. Then, the bond head 41 bonds the die D onto the package area P of the substrate S based on the image data of the substrate recognition camera 44, or bonds the die D by stacking it on top of a die that has already been bonded onto the package area P of the substrate S.

[0021] The transport unit 50 has transport claws 51 that grip and transport the substrate S, and a transport lane 52 along which the substrate S moves. The substrate S moves in the X1 direction by driving a nut (not shown) of the transport claws 51 provided on the transport lane 52 with a ball screw (not shown) provided along the transport lane 52. With this configuration, the substrate S moves from the substrate supply unit 60 along the transport lane 52 to the bonding position, and after bonding, moves to the substrate unloading unit 70 and hands the substrate S over to the substrate unloading unit 70.

[0022] The substrate supply unit 60 removes the substrate S, which has been stored in a transport jig and carried in, from the transport jig and supplies it to the transport unit 50. The substrate unloading unit 70 stores the substrate S, which has been carried in by the transport unit 50, in the transport jig.

[0023] Next, the control unit 80 will be described with reference to Fig. 3. Fig. 3 is a block diagram showing a schematic configuration of a control system of the die bonder shown in Fig. 1.

[0024] The control system 8 comprises a control unit (controller) 80, a drive unit 86, a signal unit 87, and an optical system 88. The control unit 80 is configured as a computer having a control and arithmetic unit 81 composed of a CPU (Central Processing Unit), a storage unit 82, an input / output unit 83, a bus line 84, and a power supply unit 85. The storage unit 82 has a main storage unit 82a and an auxiliary storage unit 82b. The main storage unit 82a is configured with a RAM (Random Access Memory) that stores processing programs and the like. The auxiliary storage unit 82b is configured with an HDD (Hard Disk Drive) or SSD (Solid State Drive) that stores control data, image data, and the like required for control. The processing program is a process recipe that describes processing procedures, conditions, and the like.

[0025] The input / output device 83 has a monitor 83a that displays the device status and information, a mouse 83c that operates the monitor 83a, and an image capture device 83d that captures image data from the optical system 88. The input / output device 83 further has a motor control device 83e, an I / O signal control device 83f, and a communication device 83g.

[0026] The optical system 88 includes the wafer recognition camera 14, the stage recognition camera 34, and the substrate recognition camera 44. The wafer recognition camera 14, the stage recognition camera 34, and the substrate recognition camera 44 convert light intensity and color into numerical values.

[0027] The motor control device 83e controls the driving unit 86. The driving unit 86 includes the XY table (not shown) of the wafer supply unit 10, the pickup head table 23, and the bond head table 43.

[0028] The I / O signal control device 83f takes in signals from various sensors of the signal section 87, and controls switches and volumes that control the brightness of lighting devices of the signal section 87, etc.

[0029] The communication device 83g connects the control unit 80 to a device display PC (Personal Computer) 90 present on the network.

[0030] The control / arithmetic unit 81 takes in necessary data from the memory device 82 or the input / output device 83 via the bus line 84, performs calculations, and controls the pickup head 21 etc. via the input / output device 83, or sends information to the monitor 83a etc.

[0031] For example, the control unit 80 stores image data captured by the wafer recognition camera 14, the stage recognition camera 34, and the substrate recognition camera 44 in the storage device 82 via the image capture device 83d. Using software programmed based on the stored image data, the control and calculation device 81 recognizes the positions of the die D and the package area P of the substrate S and inspects the surfaces of the die D and the substrate S. Based on the positions of the package area P of the die D and the substrate S calculated by the control and calculation device 81, the software drives the drive unit 86 via the motor control device 83e. Through this process, the positions of the die on the wafer are recognized, and the pickup head table 23 and the bond head table 43 are operated to bond the die D onto the package area P of the substrate S.

[0032] A part of the manufacturing process of a semiconductor device using the die bonder 1 (a method for manufacturing a semiconductor device) will be described with reference to Fig. 4. Fig. 4 is a flowchart showing a method for manufacturing a semiconductor device using the die bonder shown in Fig. 1. In the following description, a control unit 80 controls the operation of each unit constituting the die bonder 1.

[0033] (Wafer loading process: process S1) A wafer cassette (not shown) containing wafer rings WR is loaded into wafer cassette lifter 11. Wafer supply unit 10 removes wafer rings WR from the wafer cassette filled with wafer rings WR and carries them onto wafer holder 12.

[0034] (Substrate loading process: Process S2) The transport jig storing the substrate S is loaded into the substrate supply section 60. The substrate supply section 60 removes the substrate S from the transport jig. The transport section 50 carries the removed substrate S into the bonding section 40.

[0035] (Pickup process: process S3) After step S1, the control unit 80 moves the wafer holder 12 so that the desired die D can be picked up from the dicing tape DT. The control unit 80 photographs the die D with the wafer recognition camera 14 to obtain image data. The control unit 80 processes the image data to calculate the amount of deviation (in the X, Y, and θ directions) of the die D on the wafer holder 12 from the die position reference point of the die bonder. The control unit 80 stores the die position reference point in advance as a predetermined position of the wafer holder 12 as the initial setting of the device. The control unit 80 processes the image data to inspect the surface of the die D.

[0036] The control unit 80 peels the die D from the dicing tape DT using the peeling unit 13 and the pickup head 21. The control unit 80 adsorbs and holds the die D peeled from the dicing tape DT onto a collet 22 provided on the pickup head 21, and transports and places the die D on the intermediate stage 31.

[0037] The control unit 80 photographs the die D on the intermediate stage 31 using the stage recognition camera 34 to acquire image data. The control unit 80 processes the image data to calculate the amount of deviation (in the X, Y, and θ directions) of the die D on the intermediate stage 31 from the die position reference point of the die bonder. Note that the control unit 80 stores the die position reference point in advance at a predetermined position on the intermediate stage 31 as the initial setting of the device. The control unit 80 performs image processing on the image data to inspect the surface of the die D.

[0038] The control unit 80 returns the pickup head 21, which has transported the die D to the intermediate stage 31, to the wafer supply unit 10. The control unit 80 peels the next die D from the dicing tape DT according to the procedure described above, and thereafter peels the dies D one by one from the dicing tape DT according to the same procedure.

[0039] (Bond process: Process S4) The control unit 80 transports the substrate S to the bond stage 46 using the transport unit 50. The control unit 80 captures an image of the substrate S placed on the bond stage 46 using the substrate recognition camera 44 to obtain image data. The control unit 80 processes the image data to calculate the amount of deviation (in the X, Y, and θ directions) of the substrate S from the substrate position reference point of the die bonder 1. Note that the control unit 80 previously stores a predetermined position of the bonding unit 40 as the initial setting for the device as the substrate position reference point.

[0040] The control unit 80 corrects the suction position of the bond head 41 based on the amount of deviation of the die D on the intermediate stage 31 calculated in step S3, and suctions the die D with the collet 42. The control unit 80 causes the bond head 41, which has suctioned the die D from the intermediate stage 31, to bond the die D to a predetermined position on the substrate S supported by the bond stage 46. The control unit 80 photographs the die D bonded to the substrate S with the substrate recognition camera 44, and performs an inspection based on the image data acquired by photographing to determine whether the die D has been bonded at the desired position, etc.

[0041] The control unit 80 returns the bond head 41 that has bonded the die D to the substrate S to the intermediate stage 31. Following the procedure described above, the control unit 80 picks up the next die D from the intermediate stage 31 and bonds it to the substrate S. The control unit 80 repeats this process until the die D has been bonded to all of the package areas P of the substrate S.

[0042] (Substrate unloading process: Process S5) The control unit 80 transports the substrate S with the die D bonded thereto to the substrate unloading unit 70. The control unit 80 removes the substrate S from the transport claws 51 at the substrate unloading unit 70 and stores it in the transport jig. The transport jig storing the substrate S is unloaded from the die bonder 1.

[0043] As described above, the die D is mounted on the substrate S and is carried out from the die bonder 1. Thereafter, for example, a transport jig storing the substrate S on which the die D is mounted is transported to a wire bonding process, where the electrodes of the die D are electrically connected to the electrodes of the substrate S via Au wires or the like. Then, the substrate S is transported to a molding process, where the die D and the Au wires are sealed with molding resin (not shown), thereby completing a semiconductor package.

[0044] Next, the sensors included in the signal unit 87 will be described with reference to Fig. 5 and Fig. 6. Fig. 5 is a diagram showing the mounting positions of the sensors in this embodiment. Fig. 6 is a diagram showing the detection directions of angular velocity and acceleration of the sensors shown in Fig. 5.

[0045] 5, the bond head 41 is attached to a mover 432 of a Y table 431 that drives the bond head 41 in the Y direction. A sensor 871 is installed on the mover 432 near the bond head 41.

[0046] Sensor 871 is configured, for example, by an inertial measurement unit (IMU) including a gyro sensor capable of detecting angular velocity along three axes and an acceleration sensor capable of detecting acceleration along three axes. As shown in Fig. 6, the detection (vibration detection) directions of angular velocity and acceleration of sensor 871 are Ax: X-direction acceleration (mG), Ay: Y-direction acceleration (mG), Az: Z-direction acceleration (mG), Gx: angular velocity around the X-axis (deg / s), Gy: angular velocity around the Y-axis (deg / s), and Gz: angular velocity around the Z-axis (deg / s). The output of sensor 871 provided on Y table 431 may be only four axes, namely, Y-direction acceleration, X-axis angular velocity, Y-axis angular velocity, and Z-axis angular velocity. These four axes are the first axis in the Y direction along which bond head 41 is driven, the rotation axis around the first axis, the rotation axis around the X direction (second axis) perpendicular to the first axis, and the rotation axis around the Z direction (third axis) perpendicular to the first and second axes. This is because Y table 431 drives bond head 41 as an object in the Y direction, and vibration in the axial direction (linear direction) is detected by the Y direction acceleration. Analog data is output from sensor 871.

[0047] The diagnostic device for the die bonder 1 will be described with reference to Fig. 7 and Fig. 8. Fig. 7 is a block diagram showing an example of the configuration of a system using the diagnostic device in the embodiment. Fig. 8 is a block diagram showing an example of the configuration of the PLC shown in Fig. 7.

[0048] 7, the diagnostic device (DGN) 110 includes a diagnostic PC 120 as a computer, a programmable logic controller (PLC) 130 as a data collection device, a storage device (STD) 140, and a monitor switcher 150. The diagnostic device 110 diagnoses the die bonder 1 based on a sensor signal from the die bonder 1.

[0049] The die bonder 1 to be diagnosed includes the control unit 80 as the above-mentioned device control computer, as well as an device display PC 90 as a computer and a display PC monitor 91 as a display device. The device display PC 90 acquires production information such as process recipes and device logs (fault records, maintenance records, etc.) from the control unit 80. The device display PC 90 supplies the production information to the diagnostic PC 120 via a network 201. The device display PC 90 is connected to a monitor switch 150 via, for example, a DVI (Digital Visual Interface) cable 203a. The display PC monitor 91 is connected to the monitor switch 150 via the DVI cable 203a. The display PC monitor 91 displays the production information from the device display PC 90 via the DVI cable 203a, the monitor switch 150, and the DVI cable 203b. The display PC monitor 91 displays the diagnosis results from the diagnostic PC 120 via the DVI cable 203c, the monitor switch 150, and the DVI cable 203b. The sensor signal from the die bonder 1 is supplied to the PLC 130 via a signal line 202 .

[0050] 8, the PLC 130 includes an input unit (IN) 131, a processing unit (CPU) 132, a memory unit (MM) 133, and an output unit (OUT) 134. The input unit 131 is connected to the sensor 871 of the die bonder 1 via a signal line 202. The input unit 131 converts the analog sensor signal into sensor information as a digital signal. The processing unit 132 processes the sensor information from the input unit 131 and stores it in the memory unit 133. The output unit 134 transfers the processed sensor information to the STD 140 via FTP (File Transfer Protocol). The STD 140 saves the sensor information.

[0051] The diagnostic PC 120 acquires sensor information from the STD 140. The diagnostic PC 120 also acquires production information from the device display PC 90 of the die bonder 1 via the network 201.

[0052] The timing of diagnosing the die bonder 1 will be described with reference to Fig. 9. Fig. 9 is a flow chart for explaining the relationship between production and diagnosis in the die bonder shown in Fig. 1.

[0053] Diagnosis of the drive axes of Y table 431 etc. may be performed after production (die bonding) has finished and while waiting for the next production to start (in a production standby state). Also, during production, diagnosis may be performed while the bonding head etc. is in a standby state, such as when the die on the wafer supplied to wafer holder 12 has finished and is being replaced with the next wafer, or while the transport jig of substrate supply unit 60 is being replaced.

[0054] A diagnostic method for the die bonder 1 will be described with reference to Fig. 10. Fig. 10 is a flowchart showing a diagnostic method for the die bonder shown in Fig. 1.

[0055] For example, when diagnosing Y table 431, control unit 80 drives bond head 41 by Y table 431 from above intermediate stage 31 to above the bond point on bond stage 46, just as during production. Alternatively, control unit 80 drives bond head 41 from above the bond point on bond stage 46 to above intermediate stage 31.

[0056] (Step S11: Data acquisition and storage) The PLC 130 acquires the time-series data (waveform data) of the sensor 871 from the die bonder 1 and stores it in the STD 140.

[0057] Specifically, the PLC 130 collects waveform data (analog data) of the sensor value of the sensor 871 during diagnosis. The PLC 130 converts the collected waveform data into digital data and saves the waveform data (first waveform data) in the STD 140 in a CSV (Comma Separated Value) file format.

[0058] (Step S12: Data processing and storage) The diagnostic PC 120 acquires first waveform data from the STD 140 at a preset timing, processes the acquired first waveform data, and saves the acquired second waveform data. In the data processing, for example, the first waveform data is subjected to a fast Fourier transformation (FFT) process for each movement on each axis, averaged over a predetermined number of movements, and second waveform data is calculated based on the averaged data (averaged data). For example, the second waveform data is calculated by calculating the value of the averaged data for each frequency band using a predetermined formula. The saved second waveform data is accumulated and used for machine learning.

[0059] A specific example of data processing will be described with reference to FIGS.

[0060] The diagnostic PC 120 performs FFT processing on the first waveform data as shown in Fig. 11 to calculate FFT data as shown in Fig. 12. Here, Fig. 11 is a diagram showing an example of first waveform data acquired multiple times initially. The horizontal axis in Fig. 11 represents time, and the vertical axis represents first waveform data. Fig. 11 shows five pieces of first waveform data. Fig. 12 is a diagram showing FFT data obtained by FFT processing the first waveform data shown in Fig. 11. The horizontal axis in Fig. 12 represents frequency, and the vertical axis represents intensity. Fig. 12 shows five pieces of FFT data.

[0061] Next, the diagnostic PC 120 calculates the average value of a plurality of FFT data as shown in Fig. 12, and calculates the averaged data as shown in Fig. 13(a). Here, Fig. 13(a) is a diagram showing the averaged data obtained by calculating the average value of the FFT data shown in Fig. 12. Fig. 13(a) shows five averaged data.

[0062] Figure 13(b) shows averaged data calculated from the average value of the FFT-processed data at the time of damage. The horizontal axis in Figure 13(b) is frequency, and the vertical axis is intensity. Figure 13(b) shows five averaged data. The intensity at the time of damage shown in Figure 13(b) has a frequency band that is larger than the initial amplitude shown in Figure 13(a).

[0063] Finally, the diagnostic PC 120 calculates second waveform data as shown in Figure 14 based on the averaged data shown in Figures 13(a) and 13(b). Figure 14 is a diagram showing the second waveform data calculated based on the averaged data shown in Figures 13(a) and 13(b). The horizontal axis in Figure 14 represents the frequency band, and the vertical axis represents the calculated value. The solid line in Figure 14 represents the initial second waveform data, and the dotted line represents the second waveform data at the time of damage. In the frequency bands of 100-200 Hz and around 700-800 Hz, the calculated value at the time of damage is larger than the initial calculated value.

[0064] (Step S13: Damage diagnosis) The diagnostic PC 120 compares the second waveform data obtained by data processing in step S12 with pre-stored normal waveform data (e.g., initial second waveform data) to diagnose whether or not damage has occurred. In other words, this step diagnoses whether or not damage has occurred by comparing the waveform data acquired during diagnosis with normal waveform data measured for each individual device before production, without using machine learning. The diagnostic PC 120 displays the diagnosis results on the display PC monitor 91.

[0065] A specific example of damage diagnosis will be described with reference to Figures 15(a) and 15(b). Figure 15(a) is a diagram illustrating the threshold value for damage determination in the frequency band of 100-200 Hz. Figure 15(b) is a diagram illustrating the threshold value for damage determination in the frequency band of 700-800 Hz.

[0066] First, the diagnostic PC 120 acquires second waveform data before and after damage as shown in FIG. 14 and determines notable frequency bands (damage determination frequency bands) that exhibit characteristic changes. In FIG. 14, the notable frequency bands that exhibit characteristic changes are 100-200 Hz and 700-800 Hz. Furthermore, as shown in FIGS. 15(a) and 15(b), the diagnostic PC 120 determines in advance how much the value changes at the time of damage for each damage determination frequency band, and sets a damage determination threshold for each frequency band. For example, in the 100-200 Hz frequency band, the damage determination threshold is set to THa, which changes by more than ΔCa from the initial calculated value. In the 700-800 Hz frequency band, the damage determination threshold is set to THa, which changes by more than ΔCb from the initial calculated value.

[0067] During damage diagnosis, the diagnostic PC 120 compares the calculated value of the damage determination frequency band of the second waveform data acquired in step S12 with the thresholds (THa, THb), and determines that damage has occurred if the calculated value exceeds the thresholds.

[0068] (Step S14: Damage prediction diagnosis) Diagnostic PC 120 compares the second waveform data obtained by data processing in step S12 with the waveform model to predict the possibility (degree of progression) of damage to Y table 431. In other words, this step uses a diagnostic model created by machine learning to perform damage prediction diagnosis using the second waveform data acquired during diagnosis, and predicts the possibility (degree of progression) of damage before it occurs.

[0069] Here, the waveform model is a diagnostic model created by collecting second waveform data before and after damage in advance using a machine learning analysis tool, and having the tool learn the relationship between the characteristics of the collected second waveform data and the state of the device (damage state). The machine learning analysis tool may be provided in the diagnostic PC 120 or outside the diagnostic device 110.

[0070] In the diagnostic model, when it is determined that damage has occurred, the extent to which the characteristics of the second waveform data have changed from their initial values ​​is determined, and a threshold (damage determination value) that does not exceed this amount of change is set.

[0071] During diagnosis, the characteristics of the acquired second waveform data are compared to calculate how close it is to the threshold. Based on the results of this calculation, the degree of progression is calculated, with the damage being set at 1, for example. For example, when the degree of progression of the damage is recognized to be 0.8 (80%), parts are prepared and a repair plan is made.

[0072] According to this embodiment, at least one of the following effects (a) to (e) is obtained.

[0073] (a) Since vibrations include not only the axial direction but also the axial rotation direction, by combining and analyzing data from six axes (at least four axes), it is possible to perform an analysis that includes not only axial vibrations (linear), but also vibrations around the axis (rotation).

[0074] (b) By obtaining the diagnosis results in terms of the degree of damage, it becomes possible to roughly predict when damage will occur. Predicting when damage will prevent unplanned production stoppages by preparing replacement parts in advance and planning maintenance work in advance, thereby minimizing lost manufacturing opportunities due to long-term stoppages.

[0075] (c) Since deterioration in mounting accuracy is not detected from the mounting results but is predicted from the operating conditions, it is possible to grasp damage before it causes a large number of defects.

[0076] (d) Damage diagnosis provides results even in situations where it is difficult for humans to make a judgment, making it possible to identify damage before it has an impact on the actual product, such as on mounting accuracy.

[0077] (e) Waveform data obtained during diagnosis is saved, allowing data for machine learning to be accumulated.

[0078] The disclosure made by the present inventors has been specifically described above based on the embodiments, but the present disclosure is not limited to the above-described embodiments and can be modified in various ways.

[0079] For example, in the embodiment, the diagnosis of the Y table 431 has been described, but the present invention is not limited to this. Tables other than the Y table 431 of the bond head table 43, the pickup head table 23, and the XY table of the wafer supply unit 10 can also be diagnosed in the same way.

[0080] In the embodiment, an example in which the diagnostic device is provided outside the die bonder has been described, but this is not limited to this. Part or all of the diagnostic device may be provided inside the die bonder. For example, the diagnostic PC 120, the PLC 130, and the monitor switch 150 may be configured integrally with the device display PC 90 or the device control PC 80.

[0081] In the embodiment, an example using a die attach film has been described, but this is not limited to this. A preform unit that applies adhesive to the substrate may be provided, eliminating the need for a die attach film. The preform unit includes a preform head that applies the paste adhesive and a preform table that drives the preform head in vertical and horizontal directions. The preform table may be configured in the same manner as in the embodiment.

[0082] In the embodiment, a die bonder has been described in which a die is picked up from a wafer supply unit by a pickup head and placed on an intermediate stage, and the die placed on the intermediate stage is bonded to a substrate by a bond head, but the present invention is not limited to this and can be applied to any die bonding apparatus that picks up a die from a wafer supply unit.

[0083] For example, the present invention can be applied to a die bonder that does not have an intermediate stage and a pick-up head and that bonds a die from a wafer supply unit to a substrate with a bond head.

[0084] It can also be applied to a flip-chip bonder that does not have an intermediate stage, picks up a die from a wafer supply unit, flips the pick-up head up and down, and delivers the die to the bond head, which then bonds the die to the substrate. [Explanation of symbols]

[0085] 41 Bond head (object) 110 Diagnostic equipment 120 Diagnostic PC (computer) 130···PLC (data collector) 431···Y table (table) 871···Sensor

Claims

1. A diagnostic device for diagnosing a table having a sensor that drives an object and detects vibrations of at least four axes, comprising: a data collection device that acquires a sensor signal from the sensor for each axis during diagnosis to collect first waveform data; a computer configured to process the collected first waveform data to generate second waveform data, store the second waveform data, and compare normal waveform data measured for each individual device before production with the second waveform data to perform a damage diagnosis to determine whether or not there is damage; A diagnostic device comprising:

2. The diagnostic device of claim 1, The computer is further configured to perform a damage prediction diagnosis to predict the degree of damage to the table by comparing a diagnostic model created in advance through machine learning with the second waveform data.

3. The diagnostic device of claim 1, The four axes are a first axis in a linear direction in which the object is driven, a rotation axis around the first axis, a rotation axis around a second axis perpendicular to the first axis, and a rotation axis around a third axis perpendicular to the first axis and the second axis.

4. The diagnostic device of claim 1, The computer is configured to perform fast Fourier transform processing on the first waveform data, average the data over a predetermined number of operations, and calculate the second waveform data based on the averaged data.

5. The diagnostic device of claim 2, The computer is configured to collect the second waveform data before and after damage in advance using a machine learning analysis tool, and to learn the relationship between the characteristics of the collected second waveform data and the state of the device to create the diagnostic model.

6. The diagnostic device of claim 3, The diagnostic device includes a mounting apparatus in which the object and the table are provided, and the object is a bond head that picks up a die and places it on a substrate.

7. 4. A mounting apparatus having the diagnostic device of claim 3 therein, wherein the object is a bond head that picks up a die and places it on a substrate.

8. A method of manufacturing a semiconductor device by a semiconductor manufacturing apparatus connected to the diagnostic apparatus of claim 3, a step of loading a substrate; a bonding step of bonding a die to the substrate by a bond head, which is the object; Including, In the bonding step, the bond head is driven by the table.

Images