A method and system for low-void eutectic welding based on a uniform nitrogen gas wall
By using real-time X-ray image segmentation and gas nozzle pressure adjustment, the problem of high void ratio in chip eutectic welding was solved, achieving high-precision, low-void eutectic welding effect and improving welding reliability and efficiency.
Patent Information
- Application Number
- CN202311068161.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-08-23
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-08-23
AI Technical Summary
The high void ratio in existing eutectic bonding of chips affects device reliability and precision, and existing optimization methods are inefficient or have insignificant effects.
By acquiring real-time X-ray images of chip eutectic bonding, performing block segmentation and void detection, fitting the downward pressure value of the gas nozzle, and using an ultrasonic motor to control the inlet gas pressure of the gas nozzle, low-void eutectic bonding is achieved.
High-precision, low-void eutectic bonding was achieved, improving the reliability and efficiency of chip bonding and reducing the void rate.
Smart Images

Figure CN117219527B_ABST
Abstract
Description
Technical Field
[0001] This invention mainly relates to the field of eutectic welding technology, specifically a low-void eutectic welding method and system based on a uniform nitrogen gas wall. Background Technology
[0002] With the widespread application of semiconductor devices and components in engineering and commerce, their extensive use in fields requiring extremely high precision, such as radar, remote control and telemetry, and aerospace, places high demands on their reliability. Chip bonding, as a crucial step in the semiconductor device assembly process, has a significant impact on the reliability of semiconductor devices. Chip bonding methods include conductive bonding, non-conductive bonding, conductive adhesive bonding, eutectic bonding, and solder bonding. Among these, eutectic bonding, with its advantages of high thermal conductivity, low resistance, fast heat transfer, and high reliability, is a widely used and important method in chip bonding.
[0003] Eutectic bonding refers to the process of gently rubbing a chip against a gold-plated substrate under specific temperature and pressure conditions to remove the unstable oxide layer at the interface, causing the contact surfaces to melt and thus bonding the chips. However, due to incomplete removal of trapped air or volatile gases after the solder melts, voids can occur in eutectic bonding. Excessive void ratio significantly reduces the reliability and precision of semiconductor devices. Existing methods for optimizing void performance include improving solder chemistry, optimizing reflow soldering temperature profiles, improving substrate and component coatings, and optimizing pad and SMT stencil design. However, in practice, these methods have significant limitations, and excessively high void ratios can still occur.
[0004] Patent application CN114914164A discloses a method for eutectic bonding of microwave integrated circuit chips. The method involves cleaning a pre-formed substrate and solder pads to remove contaminants, then placing them and the chip on a hot plate surface within a eutectic furnace. Two cycles of vacuuming and nitrogen purging are performed, followed by preheating of the furnace cavity and introduction of formic acid before bonding. While this method can reduce the bonding void rate during eutectic bonding to some extent, the time required to create the appropriate gas environment significantly reduces efficiency in industrial applications.
[0005] Patent application CN112002650A discloses a process method for vacuum reflow eutectic welding. This method employs an improved vacuum reflow furnace for eutectic welding. The furnace uses two heat sources to provide the heat required for welding: one provides hot air heating, and the other provides radiant heat. Both hot air and radiant heating simultaneously provide the heat needed for eutectic welding, achieving rapid temperature rise. This method improves welding quality and efficiency, but its effect on improving the high void ratio is not significant. Summary of the Invention
[0006] The low-void eutectic welding method and system based on a uniform nitrogen wall provided by this invention solves the technical problem of high void rate in existing chip eutectic welding.
[0007] To solve the above-mentioned technical problems, the present invention proposes a low-void eutectic welding method based on a uniform nitrogen wall, comprising:
[0008] Real-time acquisition of X-ray images of chip eutectic bonding and calculation of the void ratio in the X-ray images.
[0009] After determining that the void ratio of the X-ray image does not meet the preset void ratio requirement, the X-ray image is divided into blocks according to the layout array of the eutectic welding gas nozzles to obtain the block regions.
[0010] Based on the void area and void gray value of the block region, the downpressure value of the eutectic welding gas nozzle is fitted.
[0011] The theoretical inlet gas pressure of the eutectic welding gas nozzle is obtained based on the downward pressure value.
[0012] Based on the theoretical value of the inlet gas pressure of the eutectic welding gas nozzle, the inlet gas pressure of the eutectic welding gas nozzle is adjusted to achieve low-void eutectic welding.
[0013] Furthermore, based on the layout array of the eutectic welding gas nozzles, the X-ray image is preprocessed before being segmented into blocks, specifically including:
[0014] The acquired X-ray images are subjected to image denoising and image enhancement, where the specific formula for image enhancement is as follows:
[0015]
[0016]
[0017]
[0018]
[0019] Among them, V in and V out Let A and γ represent the grayscale values of the input and output images, respectively. Let A be the grayscale scaling factor, γ be the correction parameter, D be the contrast of the input image, and P be the grayscale value. δ (m,n) represents the pixel distribution probability where the gray level difference between adjacent pixels m and n is δ, δ(m,n)=|I m -I n |,I m I ndenoted as pixel values of pixels m and n, respectively; τ is the contrast constant of the input image; V1 and V2 represent low-contrast and medium-contrast images, respectively; and σ and μ are the standard deviation and variance of the input image, respectively.
[0020] Furthermore, based on the layout array of the eutectic welding gas nozzles, the X-ray image is segmented into blocks, including:
[0021] By applying the nozzle array base profile corresponding to the layout array of eutectic welding gas nozzles to an X-ray image, three block types are obtained. The first block type contains a large nozzle action area and a 1 / 4 small nozzle action area, the second block type contains a large nozzle action area and two 1 / 4 small nozzle action areas, and the third block type contains a large nozzle action area and three 1 / 4 small nozzle action areas.
[0022] Based on three types of segmentation, X-ray images are segmented into blocks to obtain segmented regions.
[0023] Furthermore, after obtaining the segmented region, before fitting the downward pressure value of the eutectic welding gas nozzle based on the void area and void gray value of the segmented region, the following steps are also included:
[0024] Obtain the maximum area of a block-based region. Specifically, the maximum area of a block-based region refers to the area of the largest hole in the block-based region.
[0025] Based on the maximum hole area in the block, a circular convolution kernel is designed to perform adaptive circular convolution operations on X-ray images. The circular convolution kernel is represented as follows:
[0026]
[0027]
[0028]
[0029] in O i and w i S represents the circular convolution kernel, the kernel center value, and the kernel size of the i-th block region, respectively. imax This represents the maximum area of the block size in the i-th block size region.
[0030] Otsu thresholding was used to extract holes in X-ray images after adaptive circular convolution to obtain holes in block-shaped regions.
[0031] Furthermore, based on the void area and void grayscale value of the blocky region, the fitted downward pressure value of the eutectic welding gas nozzle includes:
[0032] The first fitting variable is obtained based on the average gray value of all holes in the block region.
[0033] The ratio of the area of the void in the blocky region to the area of the blocky region is calculated to obtain the second fitting variable.
[0034] Based on the type of the block-based region, a third fitting variable is obtained.
[0035] Based on the first, second, and third fitting variables, a multinomial fitting model based on the Laplace distribution is used to obtain the downpressure value of the eutectic welding gas nozzle.
[0036] Furthermore, based on the first, second, and third fitting variables, the fitting formula for obtaining the downward pressure value of the eutectic welding gas nozzle using a multinomial fitting model based on the Laplace distribution is as follows:
[0037]
[0038]
[0039] Where, p i2 Let represent the downpressure value of the i-th block region, and b1, b2, b3, μ1, μ2, μ3 be the first, second, third, fourth, fifth, and sixth fitting parameters of the multinomial fitting model based on the Laplace distribution, respectively. h i m i and c i These are the first, second, and third fitted variables for the i-th block region, respectively.
[0040] Furthermore, based on the theoretical value of the inlet gas pressure of the eutectic welding gas nozzle, the adjustment of the inlet gas pressure of the eutectic welding gas nozzle includes:
[0041] Determine whether the actual value of the inlet gas pressure of the welding gas nozzle is equal to the theoretical value. If they are equal, there is no need to adjust the inlet gas pressure of the welding gas nozzle. Otherwise, adjust the opening of the valve that controls the inlet gas pressure of the welding gas nozzle.
[0042] Furthermore, adjusting the valve opening for controlling the inlet gas pressure of the welding gas nozzle includes:
[0043] Measure the motor speed of the ultrasonic motor, which is used to control the valve opening of the welding gas nozzle inlet pressure;
[0044] Based on the actual and theoretical values of the inlet gas pressure of the welding gas nozzle, as well as the motor speed, a variable coefficient PID controller based on logistic is used to control the motor speed, so that the motor rotates the valve controlling the inlet gas pressure of the welding gas nozzle to the target opening position within a limited time.
[0045] Furthermore, the formula for calculating the PID control coefficients of the logistic-based variable coefficient PID controller is as follows:
[0046]
[0047] Where, k pv k iv k dv These are the PID control coefficients under logistic function constraints, k and k. plow k ilow k dlow The coefficients k are respectively pv k iv k dv The lower bound of the predetermined range, k phigh k ihigh k dhigh The coefficients k are respectively pv k iv k dv The upper bound of the predetermined range, e(t) is the error function, and α is the conversion coefficient used to adjust the sharpness of the conversion between the lower and upper bounds.
[0048] The low-void eutectic welding system based on a uniform nitrogen wall provided by the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the low-void eutectic welding method based on a uniform nitrogen wall provided by the present invention.
[0049] The purpose of this invention is to propose a low-void eutectic bonding method based on a uniform nitrogen gas wall, providing a feasible high-precision, low-void solution for eutectic bonding of microchips. The invention also aims to propose a block-based void detection algorithm based on real-time X-ray images of chip eutectic bonding, a pressure conversion model for block-based voidness, and a gas nozzle pressure conversion model, providing a theoretical basis for the realization of a block-based uniform nitrogen gas wall. Finally, the invention utilizes a digital signal processor to control an ultrasonic motor drive system, enabling the ultrasonic motor to control the opening of the gas valve, ensuring real-time and reliable pressure under the uniform nitrogen gas wall.
[0050] The present invention provides a solder defect identification method and system based on small target recognition. By acquiring X-ray images of chip eutectic bonding in real time and calculating the void ratio of the X-ray images, and determining that the void ratio of the X-ray images does not meet the preset void ratio requirement, the X-ray images are segmented into blocks according to the layout array of the eutectic bonding gas nozzles to obtain block regions. Based on the void area and void grayscale value of the block regions, the downward pressure value of the eutectic bonding gas nozzles is fitted. The theoretical value of the inlet gas pressure of the eutectic bonding gas nozzles is obtained based on the downward pressure value, and the inlet gas pressure of the eutectic bonding gas nozzles is adjusted based on the theoretical value of the inlet gas pressure of the eutectic bonding gas nozzles. This achieves low-voidity eutectic bonding, solves the technical problem of high void ratio in existing chip eutectic bonding, and provides a feasible high-precision low-voidity solution for eutectic bonding of microchips. Attached Figure Description
[0051] Figure 1 This is a flowchart of a low-void eutectic welding method based on a uniform nitrogen wall according to Embodiment 2 of the present invention;
[0052] Figure 2 This is a flowchart of a low-void eutectic welding method based on a uniform nitrogen wall according to Embodiment 3 of the present invention;
[0053] Figure 3 This is a diagram showing the functional domain of the eutectic welding gas nozzle array on the chip according to Embodiment 3 of the present invention;
[0054] Figure 4 This is a block diagram of the chip eutectic bonding region in Embodiment 3 of the present invention;
[0055] Figure 5 This is a cross-sectional view of the gas passage inside the nozzle in Embodiment 3 of the present invention;
[0056] Figure 6 This is a flowchart of the logistic-based variable coefficient PID ultrasonic motor control algorithm of Embodiment 3 of the present invention;
[0057] Figure 7 This is a diagram of the ultrasonic motor system according to Embodiment 3 of the present invention;
[0058] Figure 8 This is a structural block diagram of a low-void eutectic welding system based on a uniform nitrogen wall, according to an embodiment of the present invention.
[0059] Figure label:
[0060] K1, Large nozzle; K2, Small nozzle; K3, Nozzle array base; P1, Functional area of the large nozzle on the chip; P2, Functional area of the small nozzle on the chip; M1, Chip outline; M2, Nozzle array base outline; S1, First block type; S2, Second block type; S3, Third block type; L1, Pressure sensor; L2, Chip; 1, Nozzle inlet surface; 2, Nozzle jet pressure surface; 10, Memory; 20, Processor. Detailed Implementation
[0061] To facilitate understanding of the present invention, the present invention will be described more fully and in detail below with reference to the accompanying drawings and preferred embodiments, but the scope of protection of the present invention is not limited to the following specific embodiments.
[0062] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings, but the present invention can be implemented in many different ways as defined and covered by the claims.
[0063] Example 1
[0064] The low-void eutectic welding method based on a uniform nitrogen wall provided in Embodiment 1 of the present invention includes:
[0065] Step S101: Real-time acquisition of X-ray images of chip eutectic bonding and calculation of the void ratio of the X-ray images.
[0066] Step S102: After determining that the void ratio of the X-ray image does not meet the preset void ratio requirement, the X-ray image is divided into blocks according to the layout array of the eutectic welding gas nozzles to obtain the block region.
[0067] Step S103: Fit the downward pressure value of the eutectic welding gas nozzle based on the void area and void gray value of the block region.
[0068] Step S104: Obtain the theoretical inlet gas pressure value of the eutectic welding gas nozzle based on the downward pressure value.
[0069] Step S105: Adjust the inlet pressure of the eutectic welding gas nozzle according to the theoretical value of the inlet pressure of the eutectic welding gas nozzle, thereby achieving low-void eutectic welding.
[0070] The low-void eutectic welding method based on a uniform nitrogen wall provided in this invention acquires X-ray images of chip eutectic welding in real time and calculates the void ratio of the X-ray images. If the void ratio of the X-ray images does not meet a preset void ratio requirement, the X-ray images are segmented into blocks according to the layout array of the eutectic welding gas nozzles to obtain block regions. Based on the void area and void grayscale value of the block regions, the downward pressure value of the eutectic welding gas nozzles is fitted. The theoretical value of the inlet gas pressure of the eutectic welding gas nozzles is obtained based on the downward pressure value. The inlet gas pressure of the eutectic welding gas nozzles is then adjusted based on this theoretical value, thereby achieving low-void eutectic welding. This solves the technical problem of high void ratio in existing chip eutectic welding and provides a feasible high-precision, low-void solution for eutectic welding of microchips.
[0071] Example 2
[0072] like Figure 1 As shown, the low-void eutectic welding method based on a uniform nitrogen wall provided in this embodiment of the invention includes:
[0073] Step S201: Acquire X-ray images of the chip eutectic bonding in real time.
[0074] After acquiring the X-ray image in this embodiment, it is first determined whether the chip bonding is accurate. Specifically, it can be determined whether the real-time void ratio of the entire image meets the standard. If the bonding is accurate, no processing is performed and the next X-ray image is acquired; otherwise, the process proceeds to step S202.
[0075] Specifically, in this embodiment, a real-time X-ray image of chip eutectic bonding is acquired using an X-ray real-time imaging system. To address the issues of low contrast and noise interference in the X-ray image due to system equipment and external factors, image denoising and image enhancement preprocessing are performed on the acquired X-ray image.
[0076] Step S202: Perform block-based region segmentation and void detection on the X-ray image, and fit the downward pressure value of the eutectic welding gas nozzle based on the void area and void gray value of the block-based region.
[0077] Specifically, in this embodiment, the real-time X-ray image of chip eutectic bonding is segmented into blocks, and the image is convolved using a block-based adaptive circular convolution kernel. Holes are extracted from each block of the X-ray image based on OTSU threshold segmentation. The ratio of the total area of the extracted holes to the total area of the block is calculated. Based on the real-time X-ray image of chip eutectic bonding and the corresponding extracted hole regions, the downpressure value is fitted by combining the hole grayscale value with the proportion of the hole area in each block.
[0078] Step S203: Obtain the theoretical inlet gas pressure value of the eutectic welding gas nozzle based on the downward pressure value.
[0079] The theoretical value of the nozzle inlet air pressure is derived from the theoretical value of the nozzle outlet downward pressure through an air pressure conversion model. The actual value of the nozzle inlet air pressure is measured by the nozzle inlet pressure sensor. It is then determined whether the actual value of the nozzle inlet air pressure is equal to the theoretical value. If they are equal, no air pressure adjustment is required. If they are not equal, the valve opening is adjusted by ultrasonic motor drive control to change the downward pressure value.
[0080] Step S204: The variable coefficient PID (LVC-PID) ultrasonic motor control algorithm based on logistic is used to adjust the valve opening, thereby accurately bonding the chip and ultimately achieving low-void eutectic welding.
[0081] Based on the traditional PID control algorithm, a logistic function is used as the PID coefficient limiting function, allowing the PID coefficients to be continuously variable within a predetermined range, resulting in an LVC-PID ultrasonic motor controller. This controller can continuously control the speed of the ultrasonic motor, thereby controlling the opening of the gas valve. Specifically, the ultrasonic motor drive in this embodiment uses a two-phase high-frequency voltage-fed series resonant inverter, which completely converts the analog control drive system into a digital control drive system. By changing the magnitude of the DC voltage, the drive frequency is changed, thus controlling the motor speed. A 16-bit fixed-point TMS320F243 EVM digital signal processor is used to control the ultrasonic motor drive system.
[0082] Example 3
[0083] like Figure 2 As shown, the low-void eutectic welding method based on a uniform nitrogen wall provided in this embodiment of the invention includes:
[0084] Step S301: Under nitrogen atmosphere protection, acquire real-time X-ray images of chip eutectic bonding using an X-ray real-time imaging system, and simultaneously calculate the void ratio of the real-time X-ray images.
[0085] Step S302: After determining that the void ratio of the X-ray image does not meet the preset void ratio requirement, the acquired X-ray image undergoes image denoising and image enhancement preprocessing. For Gaussian noise that follows a normal distribution, Gaussian filtering is used, i.e., the original image is convolved with a two-dimensional Gaussian normal distribution function. Median filtering can be applied to the image to remove interference noise such as back holes.
[0086] The traditional expression for the gamma transform is:
[0087]
[0088] Among them, V in V represents the grayscale value of the input image. outThis represents the grayscale value of the output image, where A is the grayscale scaling factor, ranging from 0 to 1. It's easy to see that traditional gamma transform requires manual setting of the gamma parameters A and γ when adjusting the image, resulting in little variation in processing quality across different images.
[0089] This embodiment uses the input image V in The images are classified as low-contrast image V1 and medium-contrast image V2.
[0090]
[0091] Then, the γ parameters corresponding to V1 and V2 are determined according to formula (3), and the image is then enhanced:
[0092]
[0093] Among them, V in and V out Let A and γ represent the grayscale values of the input and output images, respectively. Let A be the grayscale scaling factor, γ be the correction parameter, D be the contrast of the input image, and P be the grayscale value. δ (m,n) represents the pixel distribution probability where the gray level difference between adjacent pixels m and n is δ, δ(m,n)=|I m -I n |,I m I n denoted as pixel values of pixels m and n, respectively; τ is the contrast constant of the input image; V1 and V2 represent low-contrast and medium-contrast images, respectively; and σ and μ are the standard deviation and variance of the input image, respectively.
[0094] It is easy to see that this embodiment can effectively improve the contrast of low-light areas by adaptively adjusting the gamma parameters by separating the images into low-contrast and high-contrast images, while also enhancing medium-contrast images.
[0095] Furthermore, adaptive thresholding and appropriate morphological processing are used to remove other interfering factors from the image.
[0096] Step S303: Based on the layout array of the eutectic welding gas nozzles, the X-ray image is segmented into blocks to obtain the segmented regions.
[0097] Figure 3 This is a diagram showing the area of action of the eutectic bonding gas nozzle used in this embodiment on the chip. Figure 3 It can be seen that the nozzle array base K3 of the eutectic welding gas nozzle has a large nozzle K1 and a small nozzle K2 arranged on it. When the nozzle array base contour M2 is applied to the chip contour M1 in the X-ray image, the following can be obtained: Figure 3 The scope mapping diagram is shown below. Figure 3In the diagram, P1 and P2 represent the functional areas of the large and small nozzles on the chip, respectively. By applying the nozzle array base contour corresponding to the layout array of the eutectic bonding gas nozzles to the X-ray image, the X-ray image can be segmented into three types of block regions, specifically as follows: Figure 4 The diagram shows three block types: S1 (first type), S2 (second type), and S3 (third type). It is easy to see that the first block type includes one large nozzle operating area and one quarter-small nozzle operating area; the second block type includes one large nozzle operating area and two quarter-small nozzle operating areas; and the third block type includes one large nozzle operating area and three quarter-small nozzle operating areas. This embodiment, through the design of the gas nozzle array and the block-based segmentation of the real-time X-ray image of the chip eutectic bonding, can further improve the accuracy of subsequent nitrogen wall pressure control.
[0098] Step S304: Perform adaptive circular convolution operation on the X-ray image.
[0099] In this embodiment, when performing adaptive circular convolution on X-ray images, the maximum hole area of the block region is first obtained. Specifically, the maximum hole area refers to the hole area corresponding to the largest hole in the block region.
[0100] Specifically, in this embodiment, after segmenting the X-ray image into block-like regions, morphological operations are then performed on these regions to initially segment each region into multiple holes of varying sizes. For the i-th block-like region in the image, the maximum area of the largest hole in the i-th block-like region is denoted as S. imax , which is the area of the largest hole among all holes in the i-th block region.
[0101] Then, based on the maximum hole area in the block, a circular convolution kernel is designed to perform adaptive circular convolution operations on the X-ray image. The circular convolution kernel is represented as:
[0102]
[0103]
[0104]
[0105] in O i and w i S represents the circular convolution kernel, the kernel center value, and the kernel size of the i-th block region, respectively. imax This represents the maximum area of the block size in the i-th block size region;
[0106] Otsu thresholding was used to extract holes in X-ray images after adaptive circular convolution to obtain holes in block-shaped regions.
[0107] Step S305: Fit the downward pressure value of the eutectic welding gas nozzle based on the void area and void gray value of the block region.
[0108] Specifically, in this embodiment, the first fitting variable is obtained based on the average gray value of all holes in the block region.
[0109] Then, the ratio of the area of the void in the blocky region to the area of the blocky region is calculated to obtain the second fitting variable.
[0110] Specifically, let n be the number of pixels occupied by holes in the blocky region. v If the number of pixels occupied by the block region is n, then the ratio of the area of the hole in the block region to the area of the block region is equal to the ratio of the number of pixels in the two.
[0111] Next, based on the type of the segmented region, the third fitting variable is obtained, specifically:
[0112]
[0113] Finally, based on the first, second, and third fitting variables, a multinomial fitting model based on the Laplace distribution is used to obtain the downpressure value of the eutectic welding gas nozzle.
[0114] Specifically, based on the first, second, and third fitting variables, the fitting formula for obtaining the downward pressure value of the eutectic welding gas nozzle using a multinomial fitting model based on the Laplace distribution is as follows:
[0115]
[0116] Where, p i2 Let represent the downpressure value of the i-th block region, and b1, b2, b3, μ1, μ2, μ3 be the first, second, third, fourth, fifth, and sixth fitting parameters of the multinomial fitting model based on the Laplace distribution, respectively. h i m i and c i These are the first, second, and third fitted variables for the i-th block region, respectively.
[0117] The reason why this embodiment can reduce the void ratio by controlling the downpressure is as follows: 1) Bubble removal: Bubbles are a major cause of voids. During the solidification process of liquid solder, the continuous release of gas can lead to the formation of voids. By increasing the downpressure, bubbles can escape more easily from the solder, thereby reducing the formation of voids; 2) Improved filling: Gas pressure plays an important role in the filling effect of the weld. Using an appropriate downpressure can help the solder flow better and fill the weld space. In this way, the liquid solder can more easily expel bubbles, avoiding bubbles from being trapped at the weld joint, thereby reducing the formation of voids.
[0118] This embodiment cleverly links the downpressure of the eutectic welding gas nozzle with the void ratio, offering the following advantages: 1) Multi-source information fusion: By combining real-time X-ray images of eutectic welding with void area analysis, comprehensive information about the welding process is obtained, laying the foundation for a better description of the relationship between downpressure and welding quality. 2) High intuitiveness and interpretability: By introducing visualized and quantifiable parameters such as grayscale values and void area ratio, this method provides an intuitive and interpretable way to calculate theoretical values, enabling better application in actual welding processes and maximizing the accuracy and reliability of achieving low void ratios.
[0119] Step S306: Obtain the theoretical inlet gas pressure value of the eutectic welding gas nozzle based on the downward pressure value.
[0120] Since the downward pressure of the gas nozzle used for crystal welding is mainly determined by the nozzle's inlet pressure, this embodiment uses a gas nozzle pressure conversion model to convert the downward pressure value into the theoretical value of the nozzle's inlet pressure. Figure 5 This is a cross-sectional view of the internal gas channel of the eutectic welding gas nozzle in this embodiment. L1 and L2 are the nozzle inlet pressure sensor and chip, respectively. This embodiment uses the theoretical value p of the downward-pressurized gas. i2 The theoretical value of the nozzle inlet pressure p can be derived using the pressure conversion model. i1 The derivation of the pressure conversion model is as follows:
[0121] Bernoulli's equation regarding the relationship between nozzle inlet surface 1 and nozzle jet downward pressure surface 2 is:
[0122]
[0123] Where, p i1 p represents the theoretical value of the nozzle inlet air pressure. i2 ζ represents the theoretical value of the downward pressure, v1 represents the gas flow velocity at the nozzle inlet, v2 represents the downward pressure velocity of the gas, and ζ represents the drag coefficient of the gas from the nozzle inlet surface 1 to the nozzle jet downward pressure surface 2.
[0124] The gas flow rate Q can be expressed as:
[0125] Q = S²v² (10)
[0126] Where S2 is the area of the nozzle jet downward pressure surface 2, as shown below, and r2 is the radius of the nozzle jet downward pressure surface 2.
[0127]
[0128] Combining equations (10) and (11), we can obtain:
[0129]
[0130] From equation (9), we can obtain:
[0131]
[0132] Substituting equation (13) into equation (12), we get:
[0133]
[0134] Where v1 = Q / S1, S1 is the area of nozzle inlet surface 1, and r1 is its radius, then S1 can be expressed as:
[0135] S1=πr1 2 (15)
[0136] Combining equation (14) and equation (15), we get:
[0137]
[0138] Furthermore, the theoretical value of the nozzle inlet air pressure p can be obtained. i1 , is represented as:
[0139]
[0140] The actual value of the nozzle inlet air pressure p is measured by the nozzle inlet pressure sensor L1. ir1 Determine p i1 Is it equal to p? ir1 If the values are equal, no adjustment of the air pressure is required; if they are not equal, the valve opening is adjusted by controlling the ultrasonic motor to change the downward air pressure value.
[0141] Step S307: Adjust the inlet pressure of the eutectic welding gas nozzle according to the theoretical value of the inlet gas pressure, thereby achieving low-void eutectic welding.
[0142] Specifically, when determining whether the actual value of the inlet gas pressure of the welding gas nozzle is equal to the theoretical value of the inlet gas pressure, if they are equal, there is no need to adjust the inlet gas pressure of the welding gas nozzle; otherwise, the opening of the valve controlling the inlet gas pressure of the welding gas nozzle is adjusted.
[0143] Furthermore, adjusting the valve opening for controlling the inlet gas pressure of the welding gas nozzle specifically includes:
[0144] Measure the motor speed of the ultrasonic motor, which is used to control the valve opening of the welding gas nozzle inlet pressure;
[0145] Based on the actual and theoretical values of the inlet gas pressure of the welding gas nozzle, as well as the motor speed, a variable coefficient PID controller based on logistic is used to control the motor speed, so that the motor rotates the valve controlling the inlet gas pressure of the welding gas nozzle to the target opening position within a limited time.
[0146] Specifically, the logistic function is mainly used to limit system output, and is expressed as:
[0147]
[0148] Here, parameter α is a conversion coefficient used to adjust the clarity of the transition between the lower and upper bounds. The logistic function asymptotically compresses the value of x between 0 and 1 and has the property of continuous differentiability. Therefore, the logistic function is chosen as the PID coefficient limiting function of the LVC-PID controller. The control block diagram of the LVC-PID control algorithm in this embodiment is shown in [reference needed]. Figure 6 As shown.
[0149] For the LVC-PID controller, the PID coefficients are changed using equation (19):
[0150]
[0151] Where, x low and x high These are the lower and upper bounds of the coefficients, and e(t) is the error function.
[0152] The PID coefficients are represented as follows:
[0153]
[0154] Where, k pv k iv k dv These are the PID control coefficients under logistic function constraints, k and k. plow k ilow k dlow The coefficients k are respectively pv k iv k dv The lower bound of the predetermined range, k phigh k ihigh k dhigh The coefficients k are respectivelypv k iv k dv The upper bound of the predetermined range, e(t) is the error function, and α is the conversion coefficient used to adjust the sharpness of the conversion between the lower and upper bounds.
[0155] Therefore, the transfer function of LVC-PID in this embodiment is expressed as:
[0156]
[0157] The LVC-PID controller proposed in this embodiment can provide better performance and better stability in different operating ranges.
[0158] For driving the ultrasonic motor, this embodiment employs a two-phase high-frequency voltage-fed series resonant inverter. This inverter completely converts the analog control drive system into a digital control drive system, including pulse width modulation (PWM), pulse frequency modulation (PFM), and hybrid (PWM / PFM) control. It combines a DSP with the drive system to achieve digital control. This embodiment controls the ultrasonic motor speed by changing the drive frequency f. S The specific determination method is as follows: by comparing the AC voltage V of the ultrasonic motor feedback electrode. S and fixed DC voltage V dc Determine the driving frequency f S The value of is determined. The generated signal is applied to the optocoupler, phase splitting, and voltage-controlled oscillator circuits, and then driven by the ultrasonic motor through a two-phase half-bridge series resonant inverter.
[0159] To achieve precise valve opening control, an ultrasonic motor control algorithm was designed. Based on the desired valve opening signal, and considering the time constraints of the industrial welding process, the motor speed needs to be controlled so that the motor can bring the valve plate to the specified opening position within a limited time. Specifically, when the motor receives the target valve opening signal, the control algorithm calculates the corresponding motor speed and controls the motor to achieve the target. The motor will bring the valve plate to the specified opening position via gear sets, belt drives, etc. The valve plate opening position is obtained by a positioning control device, and the motor rotation angle is obtained through an oscilloscope; the motor rotation angle corresponds to the valve plate opening position.
[0160] By changing the magnitude of the DC voltage, the switching frequency f is changed. S This controls the motor's speed. The DC voltage required for the comparison process is generated by a low-pass filter circuit. This circuit converts the PWM signal into a fixed DC voltage based on the PWM duty cycle. When the PWM signal's duty cycle changes, the generated DC voltage value changes, thereby changing the switching frequency f. S The speed of the ultrasonic motor is then adjusted to the required value.
[0161] like Figure 7 As shown, this embodiment uses a 16-bit fixed-point TMS320F243 EVM digital signal processor (DSP) to control an ultrasonic motor drive system. The DSP's event manager and general-purpose I / O (GPIO) units are used to control the ultrasonic motor. The PWM signal is generated by a general-purpose timer. The GP timer is set to continuous up-counting mode to provide an asymmetric PWM signal. Continuous wave and CCW direction signals are generated by the GPIO. The encoder signal uses the quadrature encoder pulse (QEP) circuit of the event manager. A 500-PPR encoder is used. The encoder generates two quadrature encoder signals and one index signal.
[0162] Key aspects of this invention include:
[0163] (1) Regarding the block-based void detection algorithm and the pressure conversion model of block-based void degree based on real-time X-ray images of chip eutectic welding, real-time X-ray images of chip eutectic welding are obtained through an X-ray real-time imaging system. The obtained X-ray images are preprocessed by image denoising and image enhancement. Block-based adaptive circular convolution kernel convolution operation is performed on the images. Based on OTSU threshold segmentation, voids are extracted from the X-ray images and the ratio of the total void area of each block to the block area is obtained. At the same time, based on the real-time X-ray images of chip eutectic welding and the corresponding void areas extracted, the downward pressure value is fitted by combining the void gray value with the void area ratio of each block.
[0164] (2) Regarding the gas nozzle pressure conversion model, the theoretical value of the nozzle outlet pressure p is used... i2 The theoretical value of nozzle inlet pressure p is derived through the pressure conversion model. i1 The actual value of the nozzle inlet air pressure p is measured by the nozzle inlet pressure sensor. ir1 Determine p i1 Is it equal to p? ir1 If the values are equal, no adjustment of the air pressure is required; if they are not equal, the valve opening is adjusted by controlling the ultrasonic motor to change the downward air pressure value.
[0165] (3) In the control of ultrasonic motors, based on the traditional PID control algorithm, the logistic function is used as the PID coefficient limiting function to obtain the LVC-PID controller, which is a PID controller with variable coefficients within a predetermined range. The ultrasonic motor drive adopts a two-phase high-frequency voltage-fed series resonant inverter, which is completely converted from an analog control drive system to a digital control drive system. By changing the magnitude of the DC voltage, the drive frequency is changed, and the motor speed is controlled. A 16-bit fixed-point TMS320F243 EVM digital signal processor is used to control the ultrasonic motor drive system.
[0166] Reference Figure 8 The low-void eutectic welding system based on a uniform nitrogen wall proposed in this embodiment of the invention includes a memory 10, a processor 20, and a computer program stored on the memory 10 and executable on the processor 20. When the processor 20 executes the computer program, it implements the steps of the low-void eutectic welding method based on a uniform nitrogen wall proposed in this embodiment.
[0167] The specific working process and working principle of the low-void eutectic welding system based on a uniform nitrogen wall in this embodiment can be referred to the working process and working principle of the low-void eutectic welding method based on a uniform nitrogen wall in this embodiment.
[0168] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A low-void eutectic welding method based on a uniform nitrogen gas wall, characterized in that, The method includes: Real-time acquisition of X-ray images of chip eutectic bonding and calculation of the void ratio in the X-ray images; After determining that the void ratio of the X-ray image does not meet the preset void ratio requirement, the X-ray image is divided into blocks according to the layout array of the eutectic welding gas nozzles to obtain the block region. Based on the void area and void gray value of the block region, the downpressure value of the eutectic welding gas nozzle is fitted. The theoretical inlet gas pressure of the eutectic welding gas nozzle is obtained based on the downward pressure value; Based on the theoretical value of the inlet gas pressure of the eutectic welding gas nozzle, the inlet gas pressure of the eutectic welding gas nozzle is adjusted to achieve low-void eutectic welding.
2. The low-void eutectic welding method based on a uniform nitrogen wall according to claim 1, characterized in that, Based on the layout array of the eutectic welding gas nozzles, X-ray image preprocessing is performed before segmentation, specifically including: The acquired X-ray images are subjected to image denoising and image enhancement, where the specific formula for image enhancement is as follows: Among them, V in and V out Let A and γ represent the grayscale values of the input and output images, respectively. Let A be the grayscale scaling factor, γ be the correction parameter, D be the contrast of the input image, and P be the grayscale value. δ (m,n) represents the pixel distribution probability where the gray level difference between adjacent pixels m and n is δ, δ(m,n)=|I m -I n |,I m I n denoted as pixel values of pixels m and n, respectively; τ is the contrast constant of the input image; V1 and V2 represent low-contrast and medium-contrast images, respectively; and σ and μ are the standard deviation and variance of the input image, respectively.
3. The low-void eutectic welding method based on a uniform nitrogen wall according to claim 1, characterized in that, Based on the layout array of the eutectic welding gas nozzles, the X-ray image is segmented into blocks, including: The nozzle array base contour corresponding to the layout array of the eutectic welding gas nozzles is applied to the X-ray image to obtain three block types. The first block type contains a large nozzle field and a 1 / 4 small nozzle field, the second block type contains a large nozzle field and two 1 / 4 small nozzle fields, and the third block type contains a large nozzle field and three 1 / 4 small nozzle fields. Based on three types of segmentation, X-ray images are segmented into blocks to obtain segmented regions.
4. The low-void eutectic welding method based on a uniform nitrogen wall according to claim 3, characterized in that, After obtaining the block-shaped region, the process of fitting the downpressure value of the eutectic welding gas nozzle based on the void area and void gray value of the block-shaped region also includes: Obtain the maximum void area of the block region, specifically the void area corresponding to the largest void in the block region; Based on the maximum hole area of the block, a circular convolution kernel is designed to perform adaptive circular convolution operations on X-ray images. The circular convolution kernel is represented as follows: in O i and w i S represents the circular convolution kernel, the kernel center value, and the kernel size of the i-th block region, respectively. imax This represents the maximum area of the block size in the i-th block size region; Otsu thresholding was used to extract holes in X-ray images after adaptive circular convolution to obtain holes in block-shaped regions.
5. The low-void eutectic welding method based on a uniform nitrogen wall according to any one of claims 1-4, characterized in that, Based on the void area and void gray value of the block region, the fitted downpressure values of the eutectic welding gas nozzle include: The first fitting variable is obtained based on the average gray value of all holes in the block region; Calculate the ratio of the area of the void in the blocky region to the area of the blocky region to obtain the second fitting variable; Based on the type of the block-based region, a third fitting variable is obtained; Based on the first, second, and third fitting variables, a multinomial fitting model based on the Laplace distribution is used to obtain the downpressure value of the eutectic welding gas nozzle.
6. The low-void eutectic welding method based on a uniform nitrogen wall according to claim 5, characterized in that, Based on the first, second, and third fitting variables, the fitting formula for obtaining the downward pressure value of the eutectic welding gas nozzle using a multinomial fitting model based on the Laplace distribution is as follows: Where, p i2 Let represent the downpressure value of the i-th block region, and b1, b2, b3, μ1, μ2, μ3 be the first, second, third, fourth, fifth, and sixth fitting parameters of the multinomial fitting model based on the Laplace distribution, respectively. h i m i and c i These are the first, second, and third fitted variables for the i-th block region, respectively.
7. The low-void eutectic welding method based on a uniform nitrogen wall according to claim 6, characterized in that, Based on the theoretical value of the inlet gas pressure of the eutectic welding gas nozzle, the adjustment of the inlet gas pressure of the eutectic welding gas nozzle includes: Determine whether the actual value of the inlet gas pressure of the welding gas nozzle is equal to the theoretical value. If they are equal, there is no need to adjust the inlet gas pressure of the welding gas nozzle. Otherwise, adjust the opening of the valve that controls the inlet gas pressure of the welding gas nozzle.
8. The low-void eutectic welding method based on a uniform nitrogen wall according to claim 7, characterized in that, Adjusting the valve opening for controlling the inlet gas pressure of the welding gas nozzle includes: The speed of the ultrasonic motor is measured, and the ultrasonic motor is used to control the valve opening of the welding gas nozzle inlet pressure. Based on the actual and theoretical values of the inlet gas pressure of the welding gas nozzle, as well as the motor speed, a variable coefficient PID controller based on logistic is used to control the motor speed, so that the motor rotates the valve controlling the inlet gas pressure of the welding gas nozzle to the target opening position within a limited time.
9. The low-void eutectic welding method based on a uniform nitrogen wall according to claim 8, characterized in that, The formula for calculating the PID control coefficients of a logistic-based variable coefficient PID controller is as follows: Where, k pv k iv k dv These are the PID control coefficients under logistic function constraints, k and k. plow k ilow k dlow The coefficients k are respectively pv k iv k dv The lower bound of the predetermined range, k phigh k ihigh k dhigh The coefficients k are respectively pv k iv k dv The upper bound of the predetermined range, e(t) is the error function, and α is the conversion coefficient used to adjust the sharpness of the conversion between the lower and upper bounds.
10. A low-void eutectic welding system based on a uniform nitrogen wall, the system comprising: The memory (10), the processor (20), and the computer program stored in the memory (10) and executable on the processor (20) are characterized in that the processor (20) implements the steps of the method according to any one of claims 1 to 9 when executing the computer program.
Citation Information
Patent Citations
Process method for vacuum reflow eutectic welding
CN112002650A
Eutectic welding process method for microwave integrated circuit chip
CN114914164A
Au-Sn-Ag series solder alloy, electronic component sealed using same Au-Sn-Ag series solder alloy, and electronic component-equipped device
CN105813801A
Packaging body evaluation method, analysis device and computer readable storage medium
CN116519797A
