A COB packaging device

Abstract

The application relates to a COB packaging device which comprises a data acquisition module, a target setting module, a thickness detection module, a dynamic identification module, a parameter regulation module and a verification module. The data acquisition module acquires chip heat distribution and thermal stress parameters and glue rheological parameters. The target setting module sets a differentiated target glue layer thickness of a chip center heat generation area lower than an edge stress area. The thickness detection module collects glue layer thickness data in real time through high-frequency sampling. The dynamic identification module identifies the equivalent viscosity of glue. The parameter regulation module dynamically adjusts the nozzle descending speed and the holding pressure according to the equivalent viscosity and the real-time thickness, so that the real-time thickness converges to the target thickness. The verification module adjusts the thickness of the deviation area before solidification. The device realizes closed-loop adaptive control and differentiated regulation of the glue layer thickness.

Description

Technical Field

[0001] This application relates to the field of semiconductor packaging, and more particularly to a COB packaging device. Background Technology

[0002] COB (Chip on Board) packaging technology directly mounts bare chips onto the surface of a substrate, achieving mechanical connection and heat conduction between the chip and the substrate through a die-bonding adhesive layer. It has become the mainstream packaging solution in the LED lighting and display fields. The die-bonding process is the core of COB packaging—the chip is pressed onto the substrate coated with adhesive using a die-bonding nozzle. During the pressing process, the adhesive forms a bond line thickness (BLT) that connects the chip and the substrate. The thickness of this adhesive layer directly determines the device's heat dissipation performance and long-term reliability. Currently, existing COB die-bonding equipment generally adopts an open-loop control mode of "preset parameters + fixed execution": the equipment presets the nozzle descent speed, holding pressure, and holding time based on the adhesive type and chip specifications. During the mounting process, the nozzle completes the entire process of descent, pressing, holding pressure, and lifting according to the preset parameters. After mounting, the BLT of a portion of the samples is measured offline to evaluate batch quality. Under this control mode, the industry generally regards "uniform BLT across the entire range" as the core quality evaluation indicator for the die-bonding process. Technical documents from mainstream equipment manufacturers and industry standards clearly require that the adhesive layer thickness remain consistent across the entire range. Existing encapsulation methods all aim to "ensure uniform adhesive layer thickness" and pursue consistency of BLT through adhesive self-leveling, molding and other means.

[0003] However, the aforementioned existing technologies have the following fundamental defects: First, BLT is only used as a sampling inspection indicator after the process rather than a real-time control variable. The die bonding parameters rely entirely on empirical presets and cannot respond to the dynamic changes in the actual flow state of the adhesive during the mounting process of each chip, resulting in a batch BLT dispersion of up to ±15%, which seriously restricts product consistency. Second, the existing control logic treats the chip as a rigid whole and applies uniform pressure to pursue uniform BLT across the entire area. However, this approach ignores the differentiated functional requirements of different areas of the chip for adhesive layer thickness. The core heat-generating area of ​​the chip requires a thin adhesive layer to reduce thermal resistance and improve heat dissipation efficiency, while the stress area at the edge of the chip requires a thick adhesive layer to buffer thermal cycling stress and prevent interface delamination. The industry-standard pursuit of "uniform thickness" cannot actually meet the dual requirements of thermal conductivity and stress resistance at the same time. Third, the adhesive extrusion during the die bonding process is a dynamic fluid process. The extrusion speed, pressure, and time together determine the final BLT distribution. However, existing equipment lacks the ability to perceive and actively control this dynamic process in real time and relies entirely on empirically set fixed parameters.

[0004] To address the aforementioned problems, this application provides a COB packaging device. Summary of the Invention

[0005] To address the aforementioned problems, this application provides a COB packaging device, employing the following technical solution: A COB packaging device includes a data acquisition module, a target setting module, a thickness detection module, a dynamic identification module, a parameter control module, and a verification module, wherein the data acquisition module, the target setting module, the thickness detection module, the dynamic identification module, the parameter control module, and the verification module are communicatively connected. The data acquisition module is used to acquire the thermal distribution and thermal stress parameters of the current batch of chips, as well as the rheological parameters of the current batch of adhesive. The target setting module is used to set the first target adhesive layer thickness of the central heating area and the second target adhesive layer thickness of the edge stress area of ​​each chip according to the heat distribution and thermal stress parameters, wherein the first target adhesive layer thickness is less than the second target adhesive layer thickness. The thickness detection module includes a spectral confocal thickness measurement unit integrated on the die bonding nozzle, which is used to collect real-time adhesive layer thickness data at multiple sampling points between the chip and the substrate at a sampling frequency of not less than 1kHz during the process of the die bonding nozzle driving the chip down to the substrate. The dynamic identification module is used to calculate the real-time change rate of adhesive layer thickness based on the real-time adhesive layer thickness data of each sampling point, and to identify the equivalent viscosity of adhesive under the current extrusion state in combination with the rheological parameters. The parameter control module is connected to the control end of the die bonding nozzle. It is used to adjust the descent speed and holding pressure of the die bonding nozzle in real time according to the equivalent viscosity, real-time adhesive layer thickness data and the corresponding target adhesive layer thickness. It also dynamically determines the holding pressure termination time according to the comparison results of the real-time adhesive layer thickness at each sampling point and the corresponding target adhesive layer thickness, so that the real-time adhesive layer thickness at each sampling point approaches the corresponding target adhesive layer thickness. The verification module is used to control the spectral confocal thickness measurement unit to collect the final adhesive layer thickness data of the entire chip area after the die bonding nozzle completes the pressure holding and lifts up; when the deviation between the final adhesive layer thickness of any region and the corresponding target adhesive layer thickness exceeds a preset threshold, the thickness of the deviation region is finely adjusted before the adhesive cures.

[0006] Preferably, the heat distribution and thermal stress parameters acquired by the data acquisition module include: the power level of the chip, the location and range of the chip's heat-generating core area, and the difference in the coefficient of thermal expansion between the substrate material and the chip material.

[0007] Preferably, the method by which the target setting module sets the differentiated target adhesive layer thickness is as follows: The core area of ​​chip heat generation is set as the central heat generation area, the area of ​​concentrated thermal stress at the edge of the chip is set as the edge stress area, and a gradual transition area is set between the central heat generation area and the edge stress area. The thickness of the first target adhesive layer in the central heating zone is set to 30μm to 50μm, the thickness of the second target adhesive layer in the edge stress zone is set to 70μm to 100μm, and the thickness of the target adhesive layer in the gradient transition zone smoothly increases from the thickness of the first target adhesive layer to the thickness of the second target adhesive layer.

[0008] Preferably, the spectral confocal thickness measurement unit includes one center probe and multiple edge probes; The central probe is positioned on the central axis of the die bonding nozzle and is used to collect real-time adhesive layer thickness data at the center point of the chip. The multiple edge probes are evenly distributed around the outer periphery of the die bonding nozzle to collect real-time adhesive layer thickness data at multiple sampling points on the chip edge; The measuring optical axis of each probe is parallel to the central axis of the die bonding nozzle, and the measuring end face of each probe is higher than the adsorption end face of the die bonding nozzle.

[0009] Preferably, the number of edge probes is 3 or 4.

[0010] Preferably, the dynamic identification module identifies the equivalent viscosity of the adhesive in the following specific way: Based on the change curve of the real-time adhesive layer thickness data of each sampling point over time, calculate the adhesive layer thickness change rate of the corresponding sampling point. By combining the initial viscosity and thixotropic parameters of the adhesive in the rheological parameters, the equivalent viscosity of the adhesive under the current extrusion state is inferred by using a fluid extrusion flow model.

[0011] Preferably, the specific rules for adjusting the die-bonding nozzle parameters by the parameter control module are as follows: When the equivalent viscosity of the adhesive is higher than a preset viscosity threshold, the descent speed of the die-bonding nozzle is reduced and the holding pressure is increased. When the equivalent viscosity of the adhesive is lower than a preset viscosity threshold, the descent speed of the die-bonding nozzle is increased and the holding pressure is reduced. When the equivalent viscosity of the adhesive is equal to the preset viscosity threshold, the current descent speed and holding pressure remain unchanged.

[0012] Preferably, the parameter control module is also used to control the extrusion process in stages: In the initial stage of extrusion, the die bonding nozzle is controlled to descend at a first descending speed, so that the real-time adhesive layer thickness in the heat-generating area of ​​the chip center approaches the first target adhesive layer thickness. In the later stages of extrusion, the die-bonding nozzle is controlled to descend at a second descent speed while maintaining the holding pressure. The second descent speed is less than the first descent speed, so that the adhesive flows from the center to the edge, and the real-time adhesive layer thickness in the chip edge stress zone approaches the second target adhesive layer thickness. When the real-time adhesive layer thickness in the central heating zone reaches the first target adhesive layer thickness, and the real-time adhesive layer thickness in the edge stress zone does not reach the second target adhesive layer thickness, the holding pressure is reduced and the position of the die-bonding nozzle is kept unchanged, so that the adhesive in the edge area continues to fill to the second target adhesive layer thickness under capillary action.

[0013] Preferably, the method by which the verification module performs thickness fine-tuning is as follows: When the final adhesive layer thickness in the central heating area is greater than the first target adhesive layer thickness and the deviation exceeds the preset threshold, the die bonding nozzle is controlled to apply instantaneous point pressure to the central area of ​​the chip. When the final adhesive layer thickness in the edge stress zone is less than the second target adhesive layer thickness and the deviation exceeds the preset threshold, the micro-adhesive replenishment device is controlled to replenish adhesive in the chip edge region.

[0014] Preferably, it also includes a data output module, which is communicatively connected to the verification module, for outputting the final adhesive layer thickness global distribution data of each chip to the downstream dispensing equipment, so that the downstream dispensing equipment can dynamically adjust the dispensing amount of the corresponding area according to the final adhesive layer thickness global distribution data.

[0015] In summary, this application includes at least one of the following beneficial technical effects: 1. Real-time closed-loop control of adhesive layer thickness in the die bonding process is achieved, improving batch consistency. This application integrates a spectral confocal thickness measurement unit into the die bonding nozzle to collect real-time dynamic changes in adhesive layer thickness at multiple sampling points between the chip and the substrate at a high sampling frequency of no less than 1kHz. This upgrades BLT from a post-processing sampling indicator in existing technologies to a real-time control variable. The dynamic identification module identifies the equivalent viscosity online based on the real-time adhesive layer thickness change rate and adhesive rheological parameters. The parameter control module dynamically adjusts the nozzle's descent speed and holding pressure accordingly, forming a complete closed-loop control chain of "real-time acquisition—dynamic identification—parameter control." This ensures that each chip can obtain suitable process parameters based on the actual flow state of the adhesive during the bonding process, effectively reducing BLT dispersion caused by material fluctuations such as adhesive batch differences and substrate flatness deviations, and significantly improving batch product consistency.

[0016] 2. By actively controlling the thickness of the adhesive layer with differentiation, both the heat dissipation performance and stress resistance reliability of the chip are considered. This application breaks the long-standing industry bias of uniform optimal BLT (Bipolar Thickness) technology. The target setting module sets a lower target adhesive layer thickness for the central heat-generating area and a higher target adhesive layer thickness for the edge stress area based on the chip's thermal distribution and thermal stress parameters, actively achieving a differentiated adhesive layer thickness distribution of "thinner at the center and thicker at the edges." The thinner adhesive layer at the center shortens the heat conduction path from the chip to the substrate, effectively reducing the chip junction temperature and improving heat dissipation efficiency; the thicker adhesive layer at the edges increases the adhesive volume in the stress buffer area, effectively dispersing and absorbing the interfacial shear stress caused by the mismatch in the thermal expansion coefficients of the chip and substrate materials during thermal cycling, reducing the risk of delamination failure. This differentiated control allows different areas of the same chip to obtain adhesive layer thicknesses adapted to their functional requirements, addressing both thermal conductivity and stress resistance needs from the packaging source.

[0017] 3. Adaptive compensation for material fluctuations is achieved, improving process robustness. This application achieves automatic adaptation to material fluctuations such as differences in adhesive viscosity between different batches, substrate flatness deviations, and chip thickness dispersion by identifying the equivalent viscosity of the adhesive in real time during the die bonding process and dynamically adjusting the nozzle parameters based on viscosity feedback. When the equivalent viscosity of the adhesive is too high (insufficient flowability), the system automatically reduces the nozzle descent speed and increases the holding pressure to compensate for the difficulty in discharging the adhesive; when the equivalent viscosity of the adhesive is too low (excessive flowability), the system automatically increases the nozzle descent speed and decreases the holding pressure to avoid an excessively thin adhesive layer. The entire control process requires no manual intervention, significantly improving the process's adaptability to different material states and reducing the cost of repeated parameter adjustments caused by material fluctuations.

[0018] 4. By moving the quality control checkpoint forward to before curing, batch scrapping is avoided. Under the control of the verification module, this application completes the final measurement and deviation judgment of the adhesive layer thickness across the entire chip before the adhesive cures. For areas with excessive deviations, fine-tuning is performed before curing (excess adhesive is squeezed out by instantaneous point pressure when the center is too thick, and a small amount of adhesive is added when the edges are too thin). Compared to the post-treatment remediation mode of discovering BLT defects through offline sampling inspection after curing in existing technologies, this application moves the quality control checkpoint from post-treatment sampling inspection to full inspection and real-time correction before curing, fundamentally avoiding batch scrapping caused by BLT defects and effectively reducing production costs.

[0019] 5. This application achieves data linkage between die bonding and downstream processes, providing a foundation for collaborative control of the entire process. Through a data output module, this application outputs the final adhesive layer thickness distribution data of each chip to the downstream dispensing equipment. This allows the downstream dispensing equipment to dynamically adjust the dispensing amount in corresponding areas based on the actual adhesive layer thickness distribution of each chip—reducing the dispensing amount in areas with thinner adhesive layers to avoid overflow, and increasing the dispensing amount in areas with thicker adhesive layers to compensate for height differences. This data linkage breaks away from the existing coarse linkage mode where only position coordinates are transmitted between the die bonding and dispensing processes. It achieves real-time data interaction at the parameter level between processes, providing a data foundation for collaborative control of the entire COB packaging process. Attached Figure Description

[0020] Figure 1 A system architecture block diagram of a COB packaging device in an embodiment of this application; Figure 2 This is a schematic cross-sectional view of the die-bonding nozzle in an embodiment of this application; Figure 3 This is a schematic diagram of the probe arrangement of the spectral confocal thickness measurement unit under the tilt angle of the die-bonding nozzle in the embodiments of this application.

[0021] Explanation of reference numerals in the attached figures: 100, Data acquisition module; 200, Target setting module; 300, Thickness detection module; 310, Center probe; 320, Edge probe; 400, Dynamic identification module; 500, Parameter control module; 600, Verification module; 700, Data output module; 800, Die-bonding nozzle. Detailed Implementation

[0022] The following is in conjunction with the appendix Figures 1-3 This application will be described in further detail.

[0023] This application discloses a COB packaging device. (Refer to...) Figure 1A COB packaging device includes a data acquisition module 100, a target setting module 200, a thickness detection module 300, a dynamic identification module 400, a parameter control module 500, and a verification module 600. These modules are interconnected via an industrial Ethernet or data bus. The specific data / instruction flow between modules is as follows: the output of the data acquisition module 100 is connected to the input of the target setting module 200; the output of the target setting module 200 is connected to the input of the parameter control module 500; the output of the thickness detection module 300 is connected to the inputs of both the dynamic identification module 400 and the verification module 600; the output of the dynamic identification module 400 is connected to the input of the parameter control module 500; and the output of the parameter control module 500 is connected to the control terminal of the drive mechanism of the die-bonding nozzle 800. All these modules are integrated into the control system of the COB packaging equipment. The die-bonding nozzle 800, as the physical actuator of this device, is controlled by the parameter control module 500 and is not part of any of the aforementioned functional modules. The die bonding nozzle and its control mechanism are existing technologies in this field and will not be described in detail here.

[0024] In this embodiment, the "communication connection" between modules refers to the transmission of data and instructions via wired or wireless means. As one specific implementation, the modules are integrated into the same industrial control computer, interacting via an internal data bus; as another implementation, the modules can be distributed across different controllers, communicating via CAN bus, EtherCAT, or Ethernet.

[0025] The data acquisition module 100 is used to acquire the thermal distribution and thermal stress parameters of the current batch of chips, as well as the rheological parameters of the current batch of adhesive.

[0026] Specifically, heat distribution and thermal stress parameters include the chip's power rating, the location and extent of the chip's heat-generating core area, and the difference in the coefficients of thermal expansion between the substrate and chip materials. The chip's power rating determines its heat generation during operation; a higher power rating results in a greater heat flux density in the core area, placing higher demands on heat dissipation performance. The location and extent of the chip's heat-generating core area are determined by the chip's internal circuit layout—typically located in the chip's geometric center, where the main heat-generating components are concentrated. The difference in the coefficients of thermal expansion between the substrate and chip materials (ΔCTE) determines the magnitude of thermal stress generated at the interface between the chip and substrate during temperature changes; a larger ΔCTE results in greater thermal cycling shear stress on the chip's edge areas, requiring higher reliability of the edge adhesive layer. These parameters can be obtained from chip design specifications, substrate material data sheets, and technical data sheets provided by adhesive suppliers, or they can be pre-measured offline and then entered into the system.

[0027] Rheological parameters include the initial viscosity (zero shear viscosity) and thixotropic parameters of the adhesive. The initial viscosity reflects the flow resistance of the adhesive in a static state and is a fundamental parameter determining the ease of adhesive extrusion during extrusion. The thixotropic parameter characterizes the change in adhesive viscosity with shear time and shear rate—for thixotropic adhesives, the viscosity gradually decreases under extrusion shear and gradually recovers after shearing stops. These rheological parameters can be obtained from the technical data sheets provided by the adhesive supplier or through pre-testing with an offline rheometer.

[0028] The target setting module 200 is used to set the first target adhesive layer thickness in the central heating area and the second target adhesive layer thickness in the edge stress area of ​​each chip according to the heat distribution and thermal stress parameters obtained by the data acquisition module 100, wherein the first target adhesive layer thickness is less than the second target adhesive layer thickness.

[0029] In this embodiment, the target setting module 200 sets the differentiated target adhesive layer thickness in the following specific way: First, the core heat-generating area of ​​the chip is designated as the central heat-generating area, and the area of ​​concentrated thermal stress at the chip's edge is designated as the edge stress area. A gradual transition zone is established between the central heat-generating area and the edge stress area. The central heat-generating area typically corresponds to the region where the main heat-generating components are located inside the chip, generally extending outward from the chip's geometric center to an area covering 50% to 70% of the chip area. The edge stress area corresponds to a ring-shaped region extending inward from the chip edge, approximately 100μm to 300μm in diameter. This region is where the interfacial shear stress is most concentrated during thermal cycling. The gradual transition zone is located between the central heat-generating area and the edge stress area, and within this region, the target adhesive layer thickness smoothly increases from the target value in the central heat-generating area to the target value in the edge stress area.

[0030] Secondly, the thickness of the first target adhesive layer in the central heat-generating area is set to 30μm to 50μm, and the thickness of the second target adhesive layer in the edge stress area is set to 70μm to 100μm. The above numerical ranges are based on the following: For typical COB packaged devices (chip size 1mm×1mm to 5mm×5mm, operating temperature range -40℃ to 125℃, using epoxy resin die-attach adhesive), finite element thermo-mechanical coupling simulation shows that when the adhesive layer thickness in the central heat-generating area is 30μm to 50μm, the chip junction temperature can be reduced by 5℃ to 15℃ compared to a uniform thickness (e.g., 50μm) scheme, and the heat dissipation efficiency is improved by 20% to 30%; when the adhesive layer thickness in the edge stress area is 70μm to 100μm, the interface shear stress can be reduced by 30% to 50% compared to a uniform thickness scheme, and the risk of delamination failure under thermal shock cycling is significantly reduced. The target adhesive layer thickness in the gradient transition zone increases smoothly from the first target adhesive layer thickness to the second target adhesive layer thickness. The increase can be linear, exponential, or S-shaped to ensure that no new stress concentration points are generated in the thickness transition zone.

[0031] It should be noted that the above numerical range is an optimal range based on typical COB packaging conditions. In practical applications, those skilled in the art can obtain the target thickness value suitable for specific application scenarios through conventional thermo-mechanical simulation optimization based on specific chip size, power level, adhesive type and operating temperature range.

[0032] The aforementioned thickness detection module 300 includes a spectral confocal thickness measurement unit integrated on the die-bonding nozzle, used to collect real-time adhesive layer thickness data at multiple sampling points between the chip and the substrate at a sampling frequency of not less than 1 kHz during the process of the die-bonding nozzle driving the chip down to the substrate.

[0033] Spectroscopic confocal thickness measurement is a non-contact optical precision measurement technique. Its basic principle is as follows: A white LED light source emits a beam of broad-spectrum polychromatic light. After passing through a dispersive lens, light of different wavelengths is focused at different heights along the optical axis, forming a continuous sequence of focal points. When the surface of the object being measured is at a certain focal point, the corresponding wavelength of light is efficiently reflected by the surface. The reflected light is received by a spectrometer through a confocal optical path. The spectrometer analyzes the received reflected spectrum to determine the peak wavelength of the reflected light. Based on the calibration relationship between wavelength and distance, the distance from the probe end face to the surface of the object being measured can be calculated.

[0034] The following two methods can be used to measure the thickness of the adhesive layer: Method 1 (Double-sided reflection method): The measurement beam emitted by the spectral confocal probe passes sequentially through the chip (if the chip is a transparent or semi-transparent material) and the adhesive layer, generating reflection signals on the upper surface of the chip and the upper surface of the substrate, respectively. The spectrometer receives the two reflection peaks. By analyzing the wavelength difference corresponding to the two reflection peaks and combining the refractive index of the chip material and the refractive index of the adhesive layer material, the thickness of the adhesive layer can be calculated.

[0035] Method 2 (Single-sided reflection method): The spectral confocal probe only measures the distance from the probe tip to the upper surface of the adhesive layer (i.e., the lower surface of the chip or the free surface of the adhesive layer). Since the distance from the probe tip to the upper surface of the substrate is fixed (obtained through pre-calibration), the adhesive layer thickness = fixed distance - measured distance. This embodiment preferably adopts Method 2 because it does not depend on the light transmittance of the chip material and is suitable for measuring the adhesive layer thickness of opaque chips.

[0036] Spectral confocal thickness measurement technology has advantages such as being non-contact, high precision, high sampling frequency, and insensitivity to ambient stray light and the tilt of the measured object. Commercially available spectral confocal sensors have sampling frequencies ranging from 1kHz to 10kHz, and even up to 70kHz, which fully meets the real-time acquisition requirement of no less than 1kHz in this embodiment.

[0037] Regarding the probe arrangement, please refer to... Figure 2 and Figure 3 The spectral confocal thickness measurement unit includes one central probe 310 and multiple edge probes 320.

[0038] The center probe is positioned on the central axis of the die-bonding nozzle to collect real-time adhesive layer thickness data at the chip's center point. The measuring optical axis of the center probe 310 coincides with the central axis of the die-bonding nozzle 800, and its measuring spot illuminates the geometric center point of the chip to monitor the adhesive layer thickness change in the central heat-generating area in real time.

[0039] Multiple edge probes are evenly distributed around the outer periphery of the die-bonding nozzle to collect real-time adhesive layer thickness data at multiple sampling points on the chip edge. In a preferred embodiment, the number of edge probes is three or four. When three edge probes are used, they are evenly distributed in a 120° ring around the outer periphery of the die-bonding nozzle, with the three measurement points corresponding to three directions on the chip edge (e.g., 0°, 120°, 240°), covering the main stress concentration areas on the chip edge. When four edge probes are used, they are evenly distributed in a 90° ring around the outer periphery of the die-bonding nozzle, with the four measurement points corresponding to four directions on the chip edge (e.g., 0°, 90°, 180°, 270°), providing more comprehensive coverage of the thickness distribution on the chip edge. Having three or four edge probes is the optimal choice to balance detection accuracy and engineering feasibility. Two edge probes can only cover two opposite edges of the chip and cannot obtain the complete thickness distribution of the edge stress zone. Five or more edge probes are too dense for chip edge sampling, and the limited space around the die bonding nozzle makes installation difficult and causes mutual interference.

[0040] The measuring optical axis of each probe is parallel to the central axis of the die-bonding nozzle to ensure that the measuring beam is perpendicularly incident on the adhesive surface, avoiding measurement errors caused by tilted incident light. The measuring end face of each probe is higher than the adsorption end face of the die-bonding nozzle—the measuring end face refers to the end of the probe closest to the object being measured; the adsorption end face refers to the end of the die-bonding nozzle used to adsorb the chip. This positional relationship ensures that the probe will not contact the substrate or adhesive before the chip during the die-bonding nozzle's downward pressure, ensuring the safety of non-contact measurement and preventing damage caused by physical collision between the probe and the substrate or adhesive.

[0041] Furthermore, throughout the entire process of the die-bonding nozzle lowering the chip onto the substrate, the center probe and each edge probe work synchronously, continuously acquiring adhesive layer thickness data at their respective measurement points at a sampling frequency of no less than 1kHz. The data acquired by each probe is synchronously transmitted to the controller via a multi-channel data acquisition card. The controller performs time alignment on the data from each channel, thereby obtaining the adhesive layer thickness distribution data of the entire chip area (center point + multiple edge sampling points) at the same time.

[0042] It should be noted that the above-mentioned "the central probe is set on the central axis of the die bonding nozzle" means that the probe measurement optical path is located on the central axis of the nozzle. This positional relationship does not exclude the integration method in which the probe measurement beam passes through the vacuum adsorption channel of the nozzle.

[0043] As a specific implementation of the central probe, the spectral confocal probe is coaxially integrated inside the vacuum channel of the die-bonding nozzle—that is, the probe's measurement beam passes through the vacuum channel along the central axis of the nozzle and illuminates the adhesive layer surface in the central region of the chip. In this configuration, the probe's measurement optical path and the nozzle's vacuum adsorption channel spatially overlap but functionally do not interfere with each other: the vacuum adsorption channel has an annular or side-wall opening structure, with the central region allowing the measurement beam to pass through. The two operate in parallel without obstructing the vacuum adsorption force's pickup of the chip, nor affecting the spectral confocal probe's measurement optical path. This embodiment employs this method; see specific details for reference. Figure 2 .

[0044] As another alternative implementation of the center probe, the center probe can be offset to a position adjacent to the central axis of the die-bonding nozzle (e.g., 1mm to 2mm away from the central axis). A preset geometric offset is used to correct the spatial position of the measurement data, and an algorithm is used to deduce the adhesive layer thickness at the chip's center point, thus achieving an equivalent measurement of the adhesive layer thickness in the chip's central region. Those skilled in the art will understand that neither of these methods changes the essential meaning of the positional relationship of "the center probe being positioned on the central axis of the die-bonding nozzle."

[0045] The aforementioned dynamic identification module 400 is used to calculate the real-time change rate of adhesive layer thickness based on the real-time adhesive layer thickness data of each sampling point, and to identify the equivalent viscosity of the adhesive under the current extrusion state in combination with rheological parameters.

[0046] Specifically, regarding the calculation of the adhesive layer thickness change rate, the dynamic identification module 400 receives the adhesive layer thickness data h1(t), h2(t), ..., h from each sampling point transmitted in real time by the thickness detection module 300. n (t), where subscripts 1 to n correspond to the measurement points of the center probe and each edge probe, respectively. For each sampling point, the dynamic identification module 400 calculates the rate of change of adhesive layer thickness dh / dt based on the curve of its thickness change over time. Specifically, the calculation method is as follows: numerical differentiation is performed on the continuously acquired adhesive layer thickness data, i.e., dh / dt≈[h(t+Δt)-h(t)] / Δt, where Δt is the sampling interval (equivalent to 1ms for a 1kHz sampling frequency). To improve calculation stability, the acquired data can be appropriately filtered before performing the differentiation calculation.

[0047] Specifically, regarding the identification of the equivalent viscosity of the adhesive, the dynamic identification module 400 combines the initial viscosity and thixotropic parameters of the adhesive in the rheological parameters and uses the fluid extrusion flow model to infer the equivalent viscosity of the adhesive under the current extrusion state.

[0048] The physical principle is as follows: During die bonding, the chip descends towards the substrate under the drive of the nozzle, and the adhesive is squeezed and flows within the confined space between the chip and the substrate. This process is called "squeeze flow" in fluid mechanics. The rate of change of adhesive layer thickness has a definite physical relationship with adhesive viscosity, extrusion pressure, chip area, and current adhesive layer thickness.

[0049] As a specific implementation, for Newtonian fluids or glues that can be approximated as Newtonian fluids, the Stefan extrusion flow equation can be used to describe the above physical relationship. The Stefan equation gives the condition for maintaining two parallel circular plates at a velocity... The relationship between the compressive force F required for the fluids to move closer together and the fluid viscosity η: ; Where F is the extrusion pressure (i.e. the pressure applied by the die-bonding nozzle, which can be obtained in real time by the pressure sensor of the nozzle drive mechanism), η is the adhesive viscosity, and R is the equivalent radius of the chip (determined by the chip size, which is a known parameter). h is the rate of change of adhesive layer thickness (obtained in real time by the dynamic identification module), and h is the current adhesive layer thickness (obtained in real time by the thickness detection module).

[0050] For non-Newtonian fluids (such as thixotropic epoxy resin solids), whose viscosity varies with shear rate and time, the Newtonian form of the Stefan equation cannot be directly applied. In such cases, shear-thinning fluid models such as the Carreau-Yasuda constitutive equation can be used to describe the rheological behavior of the adhesive. ; Where η0 is the initial viscosity of the glue (zero shear viscosity, provided by the rheological parameters obtained by the data acquisition module). Let be the limiting viscosity at infinite shear rate, λ be the relaxation time, and a and n be model parameters. The shear rate (which can be calculated from the velocity distribution of the extrusion flow) is given. By combining the above constitutive equation with the momentum and continuity equations of the extrusion flow, a complete physical model of the non-Newtonian fluid extrusion flow can be established. The dynamic identification module 400 substitutes the real-time acquired adhesive layer thickness change rate dh / dt, the current adhesive layer thickness h, and the known extrusion force F and chip size R into the above model to inversely deduce the equivalent viscosity η_eff of the adhesive under the current extrusion state.

[0051] It should be noted that the Stefan equation and Carreau-Yasuda model mentioned above are well-known fluid dynamics models in the field. Their specific solution process can be completed in real time on an industrial control computer using numerical calculation methods (such as the finite difference method or Newton's iteration method) without any modification to the mechanical structure of the die bonding equipment.

[0052] The parameter control module 500 is connected to the target setting module 200, the dynamic identification module 400, and the control terminal of the die bonding nozzle, respectively. It is used to adjust the descent speed and holding pressure of the die bonding nozzle in real time according to the glue equivalent viscosity identified by the dynamic identification module 400, the glue layer thickness data collected in real time by the thickness detection module 300, and the target glue layer thickness set by the target setting module 200. It also dynamically determines the holding pressure termination time according to the comparison results of the real-time glue layer thickness at each sampling point and the corresponding target glue layer thickness, so that the real-time glue layer thickness of each area of ​​the chip converges to the corresponding target glue layer thickness.

[0053] Specifically, the parameter adjustment rules based on viscosity thresholds include a preset viscosity threshold η_th in the parameter control module 500. This viscosity threshold can be preset according to the type of adhesive and process requirements, for example, it can be set as a certain percentage of the initial viscosity of the adhesive (such as 80% to 120%), or it can be optimized and determined based on historical process data.

[0054] The parameter control module 500 compares the equivalent viscosity η_eff of the adhesive identified in real time by the dynamic identification module 400 with the preset viscosity threshold η_th, and performs the following adjustments based on the comparison result: (1) When the equivalent viscosity of the adhesive η_eff is higher than the preset viscosity threshold η_th, it indicates that the current adhesive has poor flowability and high discharge resistance. At this time, the parameter control module 500 reduces the descent speed of the die bonding nozzle and increases the holding pressure. Reducing the descent speed can give the adhesive more time to be discharged, avoiding the chip from failing to descend to the target position due to untimely discharge. Increasing the holding pressure can overcome the high flow resistance of the adhesive and ensure that the adhesive is fully squeezed to the target thickness.

[0055] (2) When the equivalent viscosity of the glue η_eff is lower than the preset viscosity threshold η_th, it indicates that the current glue has good fluidity and low discharge resistance. At this time, the parameter control module 500 increases the descent speed of the die-bonding nozzle and decreases the holding pressure. Increasing the descent speed can improve production efficiency; decreasing the holding pressure can prevent low viscosity glue from being excessively discharged during the extrusion process, resulting in an excessively thin glue layer.

[0056] (3) When the equivalent viscosity of the glue η_eff is equal to the preset viscosity threshold η_th, the parameter control module 500 maintains the current descent speed and pressure unchanged.

[0057] Through the above-mentioned real-time adjustment based on viscosity threshold, the parameter control module 500 can automatically adapt to the viscosity differences between different batches of adhesive and the viscosity changes of the same batch of adhesive caused by shearing during extrusion, thereby achieving dynamic self-adaptation of the crystallization parameters.

[0058] As a further preferred embodiment, the parameter control module 500 is also used to control the extrusion process in stages. Specifically, based on the real-time changes in the thickness of the adhesive layer between the chip and the substrate, the extrusion process is divided into an initial extrusion stage, a middle and late extrusion stage, and a third stage. The control logic for each stage is as follows: (1) Initial Extrusion Stage (First Stage): The initial extrusion stage refers to the stage when the chip just begins to contact the adhesive and the adhesive layer is relatively thick. In this stage, the adhesive layer thickness is usually greater than 100μm, and the adhesive has not yet been fully spread. The parameter control module 500 controls the die bonding nozzle to descend at a first descent speed v1, so that the real-time adhesive layer thickness in the central heat-generating area of ​​the chip quickly approaches the first target adhesive layer thickness (30μm~50μm). The first descent speed v1 is relatively fast, for example, it can be 5mm / s~10mm / s, to improve production efficiency. In this stage, due to the large adhesive layer thickness, the adhesive discharge resistance is relatively small, and the fast descent speed will not cause the adhesive layer to be too thin.

[0059] The extrusion initial stage ends when the real-time adhesive layer thickness in the central heating zone of the chip drops to 50% to 70% of the initial thickness, or the real-time adhesive layer thickness in the central heating zone approaches the first target adhesive layer thickness (e.g., within 20 μm of the first target thickness).

[0060] (2) Mid-to-late stage of extrusion (second stage): After the initial extrusion stage, the mid-to-late stage of extrusion begins. In this stage, the adhesive layer thickness has been significantly reduced, and the adhesive flows from the center area of ​​the chip to the edge area. The parameter control module 500 controls the die bonding nozzle to descend at a second descent speed v2 while maintaining the holding pressure p, where the second descent speed v2 is less than the first descent speed v1 (e.g., v2 = 1 mm / s ~ 3 mm / s). Reducing the descent speed can prevent the adhesive layer in the center from becoming too thin due to excessive descent, while giving the adhesive sufficient time to flow from the center to the edge, so that the real-time adhesive layer thickness in the stress zone at the chip edge approaches the second target adhesive layer thickness (70 μm ~ 100 μm).

[0061] During this stage, the holding pressure p is dynamically adjusted by the parameter control module 500 based on the real-time identification of the glue's equivalent viscosity—increasing the pressure when the viscosity is high and decreasing the pressure when the viscosity is low.

[0062] The end condition for the middle and late stages of extrusion is: the real-time adhesive layer thickness in the heat-generating area of ​​the chip center reaches the first target adhesive layer thickness (users can set the allowable error according to the actual situation, such as allowing an error of ±3μm).

[0063] (3) Third stage: When the real-time adhesive layer thickness in the central heat-generating area of ​​the chip reaches the first target adhesive layer thickness, while the real-time adhesive layer thickness in the edge stress area of ​​the chip has not yet reached the second target adhesive layer thickness, the third stage begins.

[0064] At this stage, the parameter control module 500 reduces the holding pressure (e.g., to 30%–50% of the previous stage's holding pressure) while maintaining the die-bonding nozzle's position. At this point, the chip's central region is supported by the substrate (the central adhesive layer has reached the target thickness), while a certain gap remains between the chip's edge region and the substrate. Since the center is in place, the chip as a whole is tilted with a "low center, high edge" orientation. In this orientation, only maintaining pressure, not extrusion pressure, continues to be applied, and the adhesive continues to flow from the central region to the edge gaps under the drive of surface tension and capillary force. The adhesive layer thickness in the edge region continues to increase under capillary action until the second target adhesive layer thickness is reached.

[0065] The third stage ends when the real-time adhesive layer thickness in the chip edge stress zone reaches the second target adhesive layer thickness (with an allowable error of ±3μm). Once the real-time adhesive layer thickness at all sampling points reaches the corresponding target adhesive layer thickness, the parameter control module 500 controls the die bonding nozzle to lift, completing the placement process.

[0066] The innovation of the above-mentioned phased control logic lies in its full utilization of the physical law of "center first, edge later" during the glue extrusion process. By adjusting the descent speed and holding pressure in stages, it actively achieves a differentiated glue layer thickness distribution of "thin center and thick edge", rather than trying to eliminate this natural trend to pursue uniform thickness as existing technologies do.

[0067] The aforementioned verification module 600 is connected to the thickness detection module 300 and the parameter control module 500. It is used to control the spectral confocal thickness measurement unit to collect the final adhesive layer thickness data of the entire chip area after the die bonding nozzle completes the pressure holding and lifts up. When the deviation between the final adhesive layer thickness of any region and the corresponding target adhesive layer thickness exceeds the preset threshold, the thickness of the deviation region is finely adjusted before the adhesive cures.

[0068] Specifically, regarding the acquisition of the final adhesive layer thickness, after the parameter control module 500 controls the die-bonding nozzle to complete the pressure holding and lift it, the chip has been mounted on the substrate, but the adhesive has not yet cured (it is in a liquid or semi-solid state). At this time, the calibration module 600 controls the center probe and all edge probes of the spectral confocal thickness measurement unit to perform measurements again, acquiring the final adhesive layer thickness data of multiple sampling points across the entire chip area. The measurement method is the same as the real-time acquisition method during the mounting process—each probe measures the distance from the probe end face to the adhesive layer surface in a non-contact manner, and the adhesive layer thickness at each sampling point is calculated in combination with the pre-calibrated fixed distance.

[0069] Furthermore, the verification module 600 compares the final adhesive layer thickness at each sampling point with the corresponding target adhesive layer thickness set by the target setting module 200, and calculates the deviation value. When the deviation between the final adhesive layer thickness of any region and the corresponding target adhesive layer thickness exceeds a preset threshold, the verification module 600 controls the corresponding actuator to fine-tune the thickness of the deviation region before the adhesive cures.

[0070] The preset threshold can be set according to process requirements, for example, it can be set to ±5μm, or to ±10% of the target adhesive layer thickness. When the deviation is within the preset threshold range, the adhesive layer thickness is considered acceptable and no fine-tuning is required; when the deviation exceeds the preset threshold, the following fine-tuning operations are performed: (1) Fine-tuning of excessive thickness in the central heating area: When the final adhesive layer thickness in the central heating area is greater than the first target adhesive layer thickness and the deviation exceeds the preset threshold, it indicates that there is too much adhesive in the central area and the chip center has not descended to the target position. At this time, the verification module 600 controls the die-bonding nozzle to descend again and apply instantaneous point pressure to the central area of ​​the chip. The so-called "instantaneous point pressure" refers to the die-bonding nozzle applying point pressure to the chip center for a short time (e.g., 0.1s to 0.5s) and with a small pressure (e.g., less than 30% of the normal holding pressure) to squeeze out the excess adhesive in the central area. Since the applied pressure is instantaneous, the pressure is small and the time is short, which will not significantly affect the edge adhesive layer that has been basically formed. After the point pressure is completed, the verification module 600 controls the spectral confocal thickness measurement unit to measure the adhesive layer thickness in the central area again to confirm that the deviation has been eliminated.

[0071] (2) Thinning of the edge stress area: When the final adhesive layer thickness in the edge stress area is less than the second target adhesive layer thickness, and the deviation exceeds the preset threshold, it indicates that the adhesive in the edge area is insufficient. At this time, the verification module 600 controls the micro-volume adhesive replenishment device to replenish adhesive in the chip edge area. The micro-volume adhesive replenishment device can be a micro-dispensing valve or a micro-injection device integrated near the die bonding head. The amount of adhesive replenished is calculated based on the difference between the measured thickness and the target thickness (for example, replenishment amount = deviation value × edge perimeter × adhesive density). After the adhesive replenishment is completed, the verification module 600 controls the spectral confocal thickness measurement unit again to measure the adhesive layer thickness in the edge area to confirm that the deviation has been eliminated.

[0072] It should be noted that the above verification and fine-tuning operations are all completed before the adhesive cures—there is usually a time window of several minutes between the completion of die bonding and the chip entering the curing oven for thermal curing, which is sufficient to complete all verification and fine-tuning operations. By performing verification and fine-tuning before curing, this embodiment upgrades the quality control of the adhesive layer thickness from post-curing sampling to full inspection and real-time correction before curing, fundamentally avoiding batch scrap due to unqualified adhesive layer thickness.

[0073] As a further preferred embodiment, the COB packaging apparatus of this embodiment also includes a data output module 700. The data output module 700 is communicatively connected to the verification module 600 and is used to output the final adhesive layer thickness distribution data of each chip to the downstream dispensing equipment.

[0074] Specifically, the data output module 700 receives the final adhesive layer thickness distribution data of each chip collected by the verification module 600—including the thickness values ​​of the chip center point and each edge sampling point and their corresponding position coordinates—and transmits the above data to the control system of the downstream dispensing equipment in a preset data format (such as CSV format, JSON format or custom binary format) via industrial Ethernet, serial communication or data file sharing.

[0075] After receiving the above data, the downstream dispensing equipment dynamically adjusts the dispensing amount in the corresponding areas based on the overall distribution data of the final adhesive layer thickness for each chip: for chip areas with thinner adhesive layers (meaning a smaller distance between the chip and the substrate), the downstream dispensing equipment reduces the amount of dammed adhesive or fluorescent adhesive dispensed in that area to prevent adhesive overflow; for chip areas with thicker adhesive layers (meaning a larger distance between the chip and the substrate), the downstream dispensing equipment increases the amount of adhesive dispensed in that area to compensate for the height difference. Through the data linkage between the above processes, parameter-level real-time linkage between the die bonding process and the dispensing process is achieved, further improving the accuracy and consistency of the entire COB packaging process.

[0076] The following example, using a specific COB packaging batch, illustrates the workflow of this implementation method.

[0077] Assuming the power density of the current batch of chips is 0.5 W / mm², the chip size is 2 mm × 2 mm, and the difference in thermal expansion coefficients ΔCTE between the substrate material (Al₂O₃ ceramic) and the chip material (Si) is approximately 4.5 × 10⁻⁶. -6 / ℃. After the data acquisition module 100 acquires the above parameters, the target setting module 200 sets the following parameters: the first target adhesive layer thickness of the central heating area (a circular area with a diameter of 1.2mm) is 40μm, the second target adhesive layer thickness of the edge stress area (a ring-shaped area 200μm inward from the chip edge) is 85μm, and the target adhesive layer thickness of the gradient transition area (the area between the central heating area and the edge stress area) smoothly increases from 40μm to 85μm.

[0078] The die-bonding nozzle carries the chip down to the substrate coated with adhesive. During the descent, the thickness detection module 300 collects adhesive thickness data from the center probe and four edge probes in real time at a sampling frequency of 2kHz. The dynamic identification module 400 calculates the adhesive thickness change rate at each sampling point in real time based on the collected data, and, combined with the initial viscosity of the adhesive (12000cP) and the thixotropic index (5), identifies the current equivalent viscosity of the adhesive in real time using the Carreau-Yasuda extrusion flow model.

[0079] Assuming the dynamic identification module 400 identifies the current equivalent viscosity of the adhesive as 15000 cP, which is higher than the preset viscosity threshold (13000 cP), the parameter control module 500 accordingly reduces the descent speed of the die-bonding nozzle (from 8 mm / s to 3 mm / s) and increases the holding pressure (from 2 N to 3 N) to compensate for the insufficient fluidity of the high-viscosity adhesive.

[0080] In the initial stage of extrusion (adhesive layer thickness decreases from the initial 150μm to 80μm), the parameter control module 500 controls the nozzle to descend rapidly at a speed of 8mm / s. When the center adhesive layer thickness drops to approximately 60μm, the extrusion enters the mid-to-late stage, the nozzle speed decreases to 3mm / s, and the holding pressure is maintained at 3N. When the center adhesive layer thickness reaches 40μm, the edge adhesive layer thickness is approximately 65μm (not yet reaching the target value of 85μm), entering the third stage—the parameter control module 500 reduces the holding pressure to 1N and maintains the nozzle position unchanged, allowing the edge adhesive to continue filling under capillary action. After approximately 2 seconds, the edge adhesive layer thickness reaches 85μm. The parameter control module 500 then controls the nozzle to lift, completing the mounting process.

[0081] The calibration module 600 controls the spectral confocal thickness measurement unit to collect the final adhesive layer thickness data. The measured center thickness is 41μm (deviation +1μm, within ±5μm threshold) and the edge thickness is 83μm (deviation -2μm, within ±5μm threshold). All are qualified and no fine-tuning is required.

[0082] The data output module 700 outputs the aforementioned thickness distribution data to the downstream dispensing equipment, which then adjusts the dispensing amount based on the measured thickness distribution of each chip. This completes the die bonding and data linkage process for a single chip.

[0083] Compared with the prior art, this embodiment has the following beneficial effects: First, by integrating the spectral confocal thickness measurement unit into the die-bonding nozzle and collecting adhesive layer thickness data in real time through high-frequency sampling, BLT is upgraded from a post-inspection indicator to a real-time control variable. Combined with closed-loop feedback control, theoretically, the batch dispersion of adhesive layer thickness can be compressed from ±15% in the existing open-loop control mode to within ±5%, effectively improving batch consistency.

[0084] Secondly, by setting a differentiated target adhesive layer thickness of "thinner at the center and thicker at the edges" based on the chip's thermal distribution and thermal stress parameters, the long-standing industry bias of "uniform BLT being optimal" is broken. Under typical COB packaging conditions, compared to the uniform thickness scheme, the differentiated thickness scheme using this implementation method, according to finite element thermo-mechanical coupling simulation verification, can reduce the chip junction temperature by approximately 5°C to 15°C and the interface shear stress by approximately 30% to 50%. These effects achieve a better balance between heat dissipation performance and thermal cycling stress reliability in the packaged device, addressing both the requirements of thermal conductivity and stress resistance from the packaging source.

[0085] Third, by identifying the equivalent viscosity of the adhesive in real time during the die bonding process and dynamically adjusting the nozzle parameters based on viscosity feedback, adaptive compensation for material fluctuations such as differences in viscosity between different batches of adhesive and deviations in substrate flatness is achieved. This eliminates the need for repeated manual adjustments to process parameters, effectively improving the robustness of the process and production efficiency.

[0086] Fourth, by performing verification and fine-tuning before curing, the quality control checkpoint for adhesive layer thickness is shifted from post-curing sampling to full inspection and real-time correction before curing, which can effectively avoid batch scrap due to unqualified adhesive layer thickness and reduce production costs.

[0087] Fifth, by outputting the adhesive layer thickness distribution data of the die bonding process to the downstream dispensing equipment, parameter-level real-time linkage between the die bonding and dispensing processes is realized, providing a data foundation for the collaborative control of the entire COB packaging process and further improving the overall packaging accuracy and consistency.

[0088] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit the scope of protection of the invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on these embodiments, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art can still combine, add, delete, or otherwise adjust the features of the various embodiments of the present invention according to the circumstances without conflict or creative effort, thereby obtaining different technical solutions that do not fundamentally depart from the concept of the present invention. These technical solutions also fall within the scope of protection of the present invention.

Claims

1. A COB packaging device, characterized in that, It includes a data acquisition module, a target setting module, a thickness detection module, a dynamic identification module, a parameter control module, and a verification module, and the data acquisition module, target setting module, thickness detection module, dynamic identification module, parameter control module, and verification module are interconnected. The data acquisition module is used to acquire the thermal distribution and thermal stress parameters of the current batch of chips, as well as the rheological parameters of the current batch of adhesive. The target setting module is used to set the first target adhesive layer thickness of the central heating area and the second target adhesive layer thickness of the edge stress area of ​​each chip according to the heat distribution and thermal stress parameters, wherein the first target adhesive layer thickness is less than the second target adhesive layer thickness. The thickness detection module includes a spectral confocal thickness measurement unit integrated on the die bonding nozzle, which is used to collect real-time adhesive layer thickness data at multiple sampling points between the chip and the substrate at a sampling frequency of not less than 1kHz during the process of the die bonding nozzle driving the chip down to the substrate. The dynamic identification module is used to calculate the real-time change rate of adhesive layer thickness based on the real-time adhesive layer thickness data of each sampling point, and to identify the equivalent viscosity of adhesive under the current extrusion state in combination with the rheological parameters. The parameter control module is connected to the control end of the die bonding nozzle. It is used to adjust the descent speed and holding pressure of the die bonding nozzle in real time according to the equivalent viscosity, real-time adhesive layer thickness data and the corresponding target adhesive layer thickness. It also dynamically determines the holding pressure termination time according to the comparison results of the real-time adhesive layer thickness at each sampling point and the corresponding target adhesive layer thickness, so that the real-time adhesive layer thickness at each sampling point approaches the corresponding target adhesive layer thickness. The verification module is used to control the spectral confocal thickness measurement unit to collect the final adhesive layer thickness data of the entire chip area after the die bonding nozzle completes the pressure holding and lifts up; when the deviation between the final adhesive layer thickness of any region and the corresponding target adhesive layer thickness exceeds a preset threshold, the thickness of the deviation region is finely adjusted before the adhesive cures.

2. The COB packaging device according to claim 1, characterized in that, The heat distribution and thermal stress parameters acquired by the data acquisition module include: the power level of the chip, the location and range of the chip's heat-generating core area, and the difference in the coefficient of thermal expansion between the substrate material and the chip material.

3. The COB packaging device according to claim 1, characterized in that, The specific method by which the target setting module sets the differentiated target adhesive layer thickness is as follows: The core area of ​​chip heat generation is set as the central heat generation area, the area of ​​concentrated thermal stress at the edge of the chip is set as the edge stress area, and a gradual transition area is set between the central heat generation area and the edge stress area. The thickness of the first target adhesive layer in the central heating zone is set to 30μm to 50μm, the thickness of the second target adhesive layer in the edge stress zone is set to 70μm to 100μm, and the thickness of the target adhesive layer in the gradient transition zone smoothly increases from the thickness of the first target adhesive layer to the thickness of the second target adhesive layer.

4. The COB packaging device according to claim 1, characterized in that, The spectral confocal thickness measurement unit includes one central probe and multiple edge probes; The central probe is positioned on the central axis of the die bonding nozzle and is used to collect real-time adhesive layer thickness data at the center point of the chip. The multiple edge probes are evenly distributed around the outer periphery of the die bonding nozzle to collect real-time adhesive layer thickness data at multiple sampling points on the chip edge; The measuring optical axis of each probe is parallel to the central axis of the die bonding nozzle, and the measuring end face of each probe is higher than the adsorption end face of the die bonding nozzle.

5. A COB packaging device according to claim 4, characterized in that, The number of edge probes is 3 or 4.

6. A COB packaging device according to claim 1, characterized in that, The dynamic identification module identifies the equivalent viscosity of the adhesive in the following specific way: Based on the change curve of the real-time adhesive layer thickness data of each sampling point over time, calculate the adhesive layer thickness change rate of the corresponding sampling point. By combining the initial viscosity and thixotropic parameters of the adhesive in the rheological parameters, the equivalent viscosity of the adhesive under the current extrusion state is inferred by using a fluid extrusion flow model.

7. A COB packaging device according to claim 1, characterized in that, The specific rules for adjusting the die-bonding nozzle parameters by the parameter control module are as follows: When the equivalent viscosity of the adhesive is higher than a preset viscosity threshold, the descent speed of the die-bonding nozzle is reduced and the holding pressure is increased. When the equivalent viscosity of the adhesive is lower than a preset viscosity threshold, the descent speed of the die-bonding nozzle is increased and the holding pressure is reduced. When the equivalent viscosity of the adhesive is equal to the preset viscosity threshold, the current descent speed and holding pressure remain unchanged.

8. A COB packaging device according to claim 7, characterized in that, The parameter control module is also used to control the extrusion process in stages: In the initial stage of extrusion, the die bonding nozzle is controlled to descend at a first descending speed, so that the real-time adhesive layer thickness in the heat-generating area of ​​the chip center approaches the first target adhesive layer thickness. In the later stages of extrusion, the die-bonding nozzle is controlled to descend at a second descent speed while maintaining the holding pressure. The second descent speed is less than the first descent speed, so that the adhesive flows from the center to the edge, and the real-time adhesive layer thickness in the chip edge stress zone approaches the second target adhesive layer thickness. When the real-time adhesive layer thickness in the central heating zone reaches the first target adhesive layer thickness, and the real-time adhesive layer thickness in the edge stress zone does not reach the second target adhesive layer thickness, the holding pressure is reduced and the position of the die-bonding nozzle is kept unchanged, so that the adhesive in the edge area continues to fill to the second target adhesive layer thickness under capillary action.

9. A COB packaging device according to claim 1, characterized in that, The specific method by which the verification module performs thickness fine-tuning is as follows: When the final adhesive layer thickness in the central heating area is greater than the first target adhesive layer thickness and the deviation exceeds the preset threshold, the die bonding nozzle is controlled to apply instantaneous point pressure to the central area of ​​the chip. When the final adhesive layer thickness in the edge stress zone is less than the second target adhesive layer thickness and the deviation exceeds the preset threshold, the micro-adhesive replenishment device is controlled to replenish adhesive in the chip edge region.

10. A COB packaging device according to claim 1, characterized in that, It also includes a data output module, which is communicatively connected to the verification module, for outputting the final adhesive layer thickness global distribution data of each chip to the downstream dispensing equipment, so that the downstream dispensing equipment can dynamically adjust the dispensing amount of the corresponding area according to the final adhesive layer thickness global distribution data.

Images