A method and system for detecting the uniformity of copper thickness and line width on a PCB board.

Abstract

This invention belongs to the field of PCB board inspection technology, specifically disclosing a method and system for detecting the uniformity of copper thickness and line width on PCB boards. By dividing the PCB board surface into multiple inspection areas, copper thickness uniformity and line width uniformity are detected in each inspection area. A uniformity evaluation mechanism with two dimensions of spatial correlation and temporal continuity is constructed to realize the dynamic uniformity judgment of each inspection area during the electroplating process, which can effectively identify early anomalies. At the same time, after completing the copper thickness and line width uniformity judgment and identifying the abnormal inspection areas, the current density of the abnormal area is dynamically correlated with the copper thickness fluctuation characteristics, and the plating solution flow rate is analyzed with hysteresis correlation characteristics. This determines the main influencing factors causing non-uniformity, realizing a closed-loop analysis from anomaly detection to root cause tracing, and providing clear directional guidance for process parameter optimization.

Description

Technical Field

[0001] This invention belongs to the field of PCB board inspection technology, and specifically discloses a method and system for detecting the uniformity of copper thickness and line width on PCB boards. Background Technology

[0002] As the core carrier of electronic components, the pattern electroplating process in the manufacturing process of PCB board is a key process for forming conductive lines. It aims to thicken the copper layer on the pre-set circuit pattern through electrolytic deposition to achieve circuit conductivity and mechanical enhancement.

[0003] However, in actual electroplating processes, uneven current distribution and differences in mass transfer of the plating solution often result in significant spatial non-uniformity in the copper layer thickness and line width on the same PCB board surface. This inconsistency can lead to reliability issues such as circuit impedance deviation and localized overheating, severely impacting product yield and performance stability. Against this backdrop, the detection of copper thickness and line width uniformity on PCB boards has become a critical monitoring step to ensure electroplating quality.

[0004] Existing technologies for detecting copper thickness and linewidth uniformity generally employ offline measurement methods and rely on static criteria with fixed thresholds for quality assessment. These methods are essentially static snapshot-style evaluations of a single point in time and a localized area, failing to fully characterize the spatial heterogeneity and temporal dynamics of copper deposition behavior on PCB boards during electroplating. When uniformity defects manifest as localized gradual deviations or temporal cumulative drift, the fixed threshold criteria, lacking the ability to model spatial distribution characteristics and temporal evolution, struggle to capture early, weak non-uniformity signals, thus limiting the timeliness and diagnostic accuracy of non-uniformity risk identification.

[0005] In addition, traditional testing can usually only output qualitative conclusions of uniformity or non-uniformity, making it difficult to locate the specific spatial location of the abnormality in real time during the electroplating process, and even more difficult to identify the source of the abnormality of the dominant process parameters that cause non-uniformity. This leads to a lack of targeted process adjustments and difficulty in tracing the source of the fault. Summary of the Invention

[0006] In order to overcome the shortcomings of the existing technology, at least one problem raised in the background technology is solved.

[0007] The technical solution adopted by the present invention to solve the technical problem is as follows: The first aspect of the present invention proposes a method for detecting the uniformity of copper thickness and line width of PCB board, including the following steps: Step 1: Divide the PCB board surface into multiple detection areas and establish a position mark for each detection area.

[0008] Step 2: Use the electroplating equipment's sensors and imaging equipment to acquire the current density, plating solution flow rate, and PCB surface image corresponding to each detection area.

[0009] Step 3: Extract the copper thickness fluctuation amplitude, which represents the uniformity of copper thickness distribution, and the line width fluctuation amplitude, which represents the uniformity of line width size, from the PCB surface image of each detection area.

[0010] Step 4: Combining spatial proximity comparison and time trend analysis, determine the uniformity of copper thickness fluctuation and linewidth fluctuation in each detection area, and identify abnormal detection areas with non-uniformity risks.

[0011] Step 5: For the identified abnormal detection areas, a time-series co-occurrence correlation analysis is performed by constructing a current density sequence and a copper thickness fluctuation amplitude sequence. At the same time, a time-delay correlation analysis is performed by constructing a plating solution flow rate sequence and a linewidth fluctuation amplitude sequence, and the main influencing factors causing non-uniformity are further determined.

[0012] A second aspect of the present invention provides a PCB board copper thickness and line width uniformity detection system, comprising the following modules: a region division module: dividing the PCB board surface into multiple detection regions and establishing a position identifier for each detection region.

[0013] Data acquisition module: Utilizes sensors and imaging equipment from the electroplating equipment to acquire the current density, plating solution flow rate, and PCB surface image corresponding to each detection area.

[0014] Feature extraction module: Extracts the copper thickness fluctuation amplitude, which represents the uniformity of copper thickness distribution, and the line width fluctuation amplitude, which represents the uniformity of line width size, from the PCB surface image of each detection area.

[0015] Uniformity determination module: Combining spatial proximity comparison and time trend analysis, it determines the uniformity of copper thickness fluctuation and linewidth fluctuation in each detection area, and identifies abnormal detection areas with non-uniformity risks.

[0016] Correlation analysis module: For the identified abnormal detection areas, time-series co-occurrence correlation analysis is performed by constructing current density sequence and copper thickness fluctuation amplitude sequence. At the same time, time-delay correlation analysis is performed by constructing plating solution flow rate sequence and linewidth fluctuation amplitude sequence, and further the main influencing factors causing non-uniformity are determined.

[0017] Combining all the above technical solutions, the positive effects of this invention are as follows: 1. This invention divides the PCB board surface into multiple detection areas, and performs copper thickness uniformity and line width uniformity detection in each detection area. It constructs a uniformity evaluation mechanism with two dimensions of spatial correlation and temporal continuity, realizes dynamic uniformity judgment of each detection area during the electroplating process, can effectively identify early non-uniformity, and improve the accuracy, continuity and interpretability of the detection results.

[0018] 2. After determining the uniformity of copper thickness and linewidth and identifying abnormal detection areas, this invention dynamically correlates the current density of the abnormal areas with the copper thickness fluctuation characteristics, and simultaneously performs hysteresis correlation analysis on the plating solution flow rate and linewidth fluctuation characteristics. This allows for the identification of the main influencing factors causing non-uniformity, achieving a closed-loop analysis from anomaly detection to root cause tracing. This overcomes the limitation of traditional methods that can only output non-uniformity results without explaining why the non-uniformity exists, providing clear directional guidance for process parameter optimization. Attached Figure Description

[0019] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0020] Figure 1 This diagram illustrates the implementation steps of a method for detecting the uniformity of copper thickness and line width on a PCB board according to the present invention.

[0021] Figure 2 This is a schematic diagram of the sampling point layout in the high graphic density area of ​​this invention.

[0022] Figure 3 This is a schematic diagram of the sampling point layout in the low-graphic-density area of ​​this invention.

[0023] Figure 4 This is a module connection diagram of a PCB board copper thickness and line width uniformity detection system according to the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Example 1

[0026] See Figure 1 As shown, the present invention proposes a method for detecting the uniformity of copper thickness and line width on a PCB board, including the following steps: Step 1: Divide the PCB board surface into multiple detection areas and establish a position marker for each detection area.

[0027] Considering the significant spatial non-uniformity of the electroplating process on the PCB board surface, with different copper thickness deposition and line width control in different areas, by dividing the detection area, the continuous board surface can be decomposed into several discrete units, which makes it easier to independently extract the local copper thickness distribution uniformity and line width distribution uniformity in each unit, avoiding information masking caused by global averaging.

[0028] The specific implementation process of the above steps is as follows: Based on the topological characteristics of the graphic layout of the PCB board, the entire board is functionally partitioned into high graphic density areas and low graphic density areas.

[0029] Understandably, PCB design involves non-uniform circuit layouts based on functional requirements. This non-uniformity directly leads to spatial differences in pattern density, specifically high pattern density areas typically appearing in chip packaging areas and areas with dense high-speed signal traces. These areas have complex wiring, small conductor spacing, a large number of copper conductors and pads per unit area, and concentrated current paths.

[0030] Low pattern density areas are commonly found in large copper gaps in power supplies, blank areas at the edges of boards, and heat dissipation isolation areas. These areas have sparse conductors and less pattern coverage per unit area.

[0031] The spatial difference in pattern density causes significant uneven current distribution during electroplating, resulting in a region-specific copper layer thickness distribution. This phenomenon provides prior knowledge for subsequent copper thickness uniformity detection.

[0032] Specifically, this manifests in the following ways: the outer conductors of densely patterned areas create an electric field shielding for the internal circuitry, making it difficult for current to effectively flow into the center of the area or into narrow gaps. Simultaneously, the limited ion diffusion path reduces the mass transfer rate, leading to enhanced local polarization. This results in the actual deposition current density being lower than the set value, thus causing uneven copper layer deposition.

[0033] In low-density areas of the pattern, the conductor distribution is sparse and the current path is simple, resulting in favorable conditions for copper deposition. However, in isolated traces or at the edges of large copper areas, the electric field lines are highly concentrated, leading to a much higher current density at the edges than at the center, which causes preferential deposition at the edges. At the same time, the overall flow rate of the plating solution is high in open areas, but the diffusion boundary layer close to the cathode surface is still relatively thick, which restricts the transfer of copper to the surface and easily forms eddies or stagnant areas at the edges or corners, further aggravating local uneven deposition. Therefore, local uneven copper deposition is prone to occur in low-density areas of the pattern.

[0034] Furthermore, the functional partitioning is not based on a specific boundary line, but is determined based on the topological features of the graphic layout. Specifically, the PCB light pattern is first converted into a binary image, then connected components are extracted using image processing techniques, and finally clustering is performed based on local pixel density to identify high graphic density areas and low graphic density areas.

[0035] Within each functional area, grid cells are divided to generate several detection regions. The grid division is constrained by the boundaries of the functional areas to ensure that each detection region is completely contained within a single functional area and does not cross the boundaries of different areas.

[0036] All generated detection regions are assigned spatial location identifiers, which contain region type codes and grid coordinate information.

[0037] This invention independently divides the functional areas based on the topological features of the graphic layout, which is beneficial for adopting differentiated sampling strategies for different regions in the detection of copper thickness distribution uniformity. The functional area boundary is used as a constraint to clearly express the regional closure of the detection unit, avoiding the feature aliasing problem caused by cross-region units, thereby ensuring the physical consistency of the extracted copper thickness distribution features.

[0038] Step 2: Use the electroplating equipment's sensors and imaging equipment to acquire the current density, plating solution flow rate, and PCB surface image corresponding to each detection area.

[0039] As a preferred implementation of the above scheme, the specific detection process for current density and plating solution flow rate is as follows: A sensor array is set up in the electroplating tank to collect current density and plating solution flow rate.

[0040] The sensors mentioned in this invention include current density sensors and plating solution flow rate sensors.

[0041] Among them, the current density sensor is used to measure the surface current density of a local area, reflecting the intensity of the electroplating current passing through a unit area.

[0042] A plating solution flow rate sensor is used to measure the local flow rate of the plating solution at a specific location.

[0043] Establish a mapping relationship between the sensor acquisition location and the detection area location marker on the PCB board, thereby assigning a corresponding current density and plating solution flow rate to each detection area.

[0044] As an optional implementation of the above operations, the following mapping and data allocation process is provided: SO1. Establish a global Cartesian coordinate system on the PCB board surface, with the upper left corner of the board surface as the origin, the horizontal rightward direction as the positive x-axis, and the vertical downward direction as the positive y-axis, constructing a unified spatial reference frame. Under this coordinate system, accurately calibrate the physical installation position of each sensor in the electroplating tank, as well as the geometric center coordinates of each detection area, ensuring that all physical entities have calculable spatial positioning.

[0045] SO2. Compare the geometric center coordinates of each detection area with the spatial distance of all sensors. If there is a unique nearest neighbor sensor, the current density and plating solution flow rate data collected by that sensor are directly assigned to that area.

[0046] If the detection area is within the coverage of multiple sensors, inverse distance weighted interpolation is used for fusion calculation.

[0047] The fusion calculation formula, taking the plating solution flow rate as an example, is as follows: ,in In the formula Indicates the first The weights of the data measured by the sensor Indicates the first The flow rate of the plating solution measured by the sensor, This indicates the sensor number within the detection area. , Indicates the first The Euclidean distance between the sensor and the geometric center of the detection area is determined by using... The reverse weighting method ensures that the data measured by the sensor that is closer to the detection area contributes more, reflecting the principle that measurements from similar locations have a stronger correlation.

[0048] Considering that weights are inversely proportional to distance, when a sensor coincides with or is extremely close to the center of the detection area, the denominator tends to approach zero, causing numerical overflow or weight distortion. Therefore, a minimum distance threshold can be set. For example, for 0.5mm, the optimized weight expression is: This ensures that the calculation remains stable even when the distance is very small, while strictly maintaining the inverse relationship that the closer the distance, the greater the weight.

[0049] SO3. Each detection area obtains a current density sequence and a plating solution flow rate sequence associated with its unique location identifier, which serve as the basis for subsequent spatiotemporal alignment and dynamic correlation analysis with copper thickness and linewidth distribution uniformity.

[0050] Step 3: Extract the copper thickness fluctuation amplitude, which represents the uniformity of copper thickness distribution, and the line width fluctuation amplitude, which represents the uniformity of line width size, from the PCB surface image of each detection area.

[0051] Specifically, the copper thickness fluctuation range includes the following extraction process: Based on the PCB surface image of each detection area, identify the functional area to which it belongs, and arrange sampling points according to the functional area as follows: High graphic density area: Determine the midpoint of the upper and lower lines in the detection area and connect them to form a central axis. This central axis and the boundary of the detection area together divide the area into four sub-areas; Sampling points are arranged at the geometric center point of the detection area, the center points of the four sub-areas, and the boundary points.

[0052] For the above scheme, the sampling point layout in high-graphic-density areas is as follows: Figure 2 .

[0053] To accurately assess the uniformity of copper thickness deposition on the board surface, structured sampling points need to be laid out within the detection area. Spatially discrete copper thickness values ​​are obtained through multi-point measurements to quantify the copper thickness fluctuation range, which serves as a uniformity evaluation index.

[0054] In areas with high pattern density, the dense conductor layout and complex electric field distribution can easily lead to local anomalies such as shielding effects and current concentration, resulting in high-frequency and strong gradient variations in deposition behavior in space. Therefore, a high-density sampling strategy is required.

[0055] This invention deploys sampling points at the geometric center of the detection area, the center points of four sub-regions, and boundary points such as corner points. The center point represents the overall level, the center point of the sub-region represents the internal heterogeneity, and the boundary points monitor the deposition deviation caused by the concentration of edge currents, representing the extreme values ​​under the influence of boundary effects. This multi-scale, full-coverage monitoring method achieves high-density sampling.

[0056] Low graphic density area: Sampling points are placed at the geometric center point and the midpoint of the four sides of each detection area.

[0057] For the above scheme, the sampling point layout in low-graphic-density areas is as follows: Figure 3 .

[0058] In areas with low pattern density, the deposition trend is relatively gentle and the main anomalies are concentrated at the edges, which can be effectively monitored by sampling key feature points.

[0059] This invention arranges sampling points at the geometric center point and the midpoints of the four sides of the detection area to form a cross-shaped distribution. The center point reflects the average deposition level of the area, while the midpoints of the sides directly capture the preferential deposition caused by the edge effect. This method controls the sampling density while ensuring spatial representativeness and improving detection efficiency.

[0060] Image grayscale values ​​are extracted from each sampling point deployed in each detection area.

[0061] Based on the pre-determined correspondence between grayscale values ​​and copper thickness through experiments, the grayscale value of each sampling point is converted into a copper thickness value.

[0062] It should be noted that the relationship between grayscale value and copper thickness is based on the physical correlation between optical reflectance characteristics and the thickness of the metal film layer. Specifically, the thickness of the copper layer on the PCB surface directly affects its optical reflectivity, which in turn affects the grayscale value of the image captured by the imaging system. The thicker the copper layer, the more stable the surface oxidation, the higher the reflectivity, and the higher the pixel grayscale value during imaging; the thinner the copper layer, the stronger the influence of the substrate, or the increased scattering due to microstructure, the lower the reflectivity, and the lower the grayscale value. Within a certain thickness range, such as 5–35 μm, there is a correlation between grayscale value and copper thickness value. This correlation can be calibrated experimentally and described by establishing a mapping relationship.

[0063] The steps for calibrating the relationship between grayscale values ​​and copper thickness using standard sample experiments are as follows: First, design and fabricate a set of standard test boards with known copper thicknesses, where the copper thickness gradient covers the target detection range, such as 10, 15, 20, 25, and 30 μm. All samples maintain the same graphic characteristics, surface treatment process, and cleanliness to ensure the uniqueness of the variables.

[0064] Secondly, using the same imaging system as the actual test, surface images of each standard sample were captured under constant ambient light, and the average gray value of each region was extracted.

[0065] Next, the measured grayscale value is paired with the corresponding actual copper thickness, which can be accurately measured offline using methods such as XRF and metallographic sectioning.

[0066] Finally, by using the data fitting function relationship between grayscale values ​​and actual copper thickness, the calibration results are stored as a correspondence between grayscale values ​​and copper thickness.

[0067] The range of copper thickness values ​​at all sampling points within each detection area is calculated as the copper thickness fluctuation range of that detection area. This parameter directly quantifies the absolute difference between the thickest and thinnest points within the same detection area, reflecting the degree of extreme deviation in the deposition process. The larger the difference, the more inconsistent the copper layer distribution in the lateral space and the worse the uniformity.

[0068] More specifically, the linewidth fluctuation range includes the following extraction process: edge detection is performed on the PCB surface image of each detection area to identify the outline of the circuit pattern.

[0069] Measurement points are set up along the vertical direction of the line contour. At each measurement point, a normal section perpendicular to the tangent direction of the contour is constructed. The line width value at that position is obtained by obtaining the Euclidean distance between corresponding points on both sides.

[0070] The range of all linewidth values ​​within each detection area is calculated as the linewidth fluctuation range of that detection area. This parameter quantifies the absolute difference between the widest and narrowest linewidths among all measurement points within the detection area.

[0071] Step 4: Combining spatial proximity comparison and time trend analysis, determine the uniformity of copper thickness fluctuation and linewidth fluctuation in each detection area, and identify abnormal detection areas with non-uniformity risks.

[0072] The specific implementation of the above steps is as follows: For each detection area, all detection areas adjacent to its boundary are identified as neighboring areas.

[0073] The copper thickness and linewidth fluctuations of each detection area were continuously recorded at different time points.

[0074] The copper thickness fluctuation range and line width fluctuation range of each detection area are compared with the average copper thickness fluctuation range of its neighboring areas.

[0075] The copper thickness fluctuation range and linewidth fluctuation range of each detection area are compared with the copper thickness fluctuation range and linewidth fluctuation range of the previous sampling time.

[0076] If either the copper thickness fluctuation or the linewidth fluctuation in a certain detection area is higher than the average fluctuation of neighboring areas, or if either the copper thickness fluctuation or the linewidth fluctuation shows an upward trend over multiple consecutive sampling periods, then the detection area is identified as an abnormal detection area.

[0077] In the specific implementation of the above scheme, the upward trend can be determined by collecting data on the copper thickness fluctuation amplitude and line width fluctuation amplitude at multiple consecutive sampling times, such as the last 5 time points, to construct a time-series fluctuation curve. Based on this curve, it can be identified whether it is monotonically increasing. If it is monotonically increasing, it is determined that there is an upward trend.

[0078] It is important to understand that the process field in PCB electroplating exhibits significant spatial continuity and neighborhood correlation. Therefore, under normal operating conditions, the fluctuation range of copper thickness and linewidth in adjacent detection areas should be similar, demonstrating good spatial consistency. Furthermore, the electroplating process is essentially a gradual material deposition process, and under normal circumstances, its fluctuation range should remain stable over time or only undergo a slow, smooth evolution.

[0079] When the fluctuation range of copper thickness or linewidth in a certain detection area is significantly higher than the average of its neighboring areas, or shows an upward trend at multiple consecutive sampling times, it indicates that the uniformity degradation of the area has deviated from the normal evolution path, reflecting the local abruptness of non-uniformity in space or the accelerated evolution in time. Compared with the traditional anomaly identification method that relies on a preset fixed threshold, this method realizes the adaptive generation of the judgment threshold, which significantly improves the accuracy of anomaly identification.

[0080] Step 5: For the identified abnormal detection areas, a time-series co-occurrence correlation analysis is performed by constructing a current density sequence and a copper thickness fluctuation amplitude sequence. At the same time, a time-delay correlation analysis is performed by constructing a plating solution flow rate sequence and a linewidth fluctuation amplitude sequence, and the main influencing factors causing non-uniformity are further determined.

[0081] As a means to achieve the above scheme, the time-series co-occurrence correlation analysis process of current density and copper thickness fluctuation amplitude is as follows: obtain the copper thickness fluctuation amplitude sequence and the corresponding current density sequence collected at multiple consecutive time points in the same detection area during the observation period.

[0082] The copper thickness fluctuation amplitude sequence and the current density sequence were used to construct copper thickness fluctuation time curves and current density time curves, respectively.

[0083] Capture the inflection point of the copper thickness fluctuation time curve and extract the rising point of the copper thickness fluctuation.

[0084] Inflection points are also captured on the current density-time curve to extract the points where the current density decreases.

[0085] It should be explained that the rising point of copper thickness fluctuation indicates an increase in lateral copper thickness non-uniformity, while the falling point of current density indicates a weakening of the electrochemical driving force at the cathode surface.

[0086] By performing time-series alignment and matching analysis on the two types of feature points, a set of event pairs of copper thickness fluctuation increase and current density decrease is constructed.

[0087] When the current density of a certain area on a PCB board decreases due to poor contact, anode loss, or local shielding, the copper ion reduction rate decreases accordingly, resulting in a reduction in the amount of copper deposited. Meanwhile, adjacent areas maintain a higher current density and deposition rate, causing the relative deposition rate difference between different locations within that area to widen. This leads to an increase in the dispersion of local copper thickness distribution, manifested as a significant increase in the fluctuation range of copper thickness.

[0088] The above process reveals that current density is the dominant influencing mechanism of copper thickness uniformity, and there is a clear physical causal relationship between the two. Based on this, the copper thickness fluctuation rise point and the current density fall point are captured in the time curve to form the copper thickness fluctuation rise-current density fall event, which can realize the dynamic characterization and time sequence locking of this causal relationship.

[0089] Because the process has dynamic time-varying and intermittent disturbance characteristics, the evolution of copper thickness fluctuation amplitude is non-monotonic and often shows a multi-peak undulating time response. Therefore, multiple fluctuation rise points can be identified, thus forming multiple pairs of copper thickness fluctuation rise-current density decrease events.

[0090] If the time interval between two types of feature points in the set falls within the correlation window for a certain copper thickness fluctuation increase-current density decrease event, it is determined to be a time-series co-occurrence event.

[0091] Applying this to the above scheme, given that current drop usually precedes copper thickness unevenness, the current density drop event often precedes the copper thickness fluctuation rise event. There is a certain response delay between these two events. By setting an association window such as 1.5 times the sampling time interval, it can be determined whether the two events co-occur in time.

[0092] The frequency of time-series co-occurrence events occurring in the anomaly detection area during the observation period is statistically analyzed, and the co-occurrence rate is calculated. The co-occurrence rate can be defined as the ratio of the frequency of time-series co-occurrence events to the total number of events where copper thickness fluctuation increases and current density decreases. If the co-occurrence rate reaches the co-occurrence judgment value, it indicates that there is a dynamic correlation between the deterioration of copper thickness uniformity and the decrease in local current density in the anomaly detection area. Therefore, it is determined that there is a strong correlation between copper thickness uniformity and current density in the detection area, indicating that the copper thickness fluctuation in this area is mainly driven by uneven current distribution.

[0093] The co-occurrence determination value is set above to eliminate random fluctuation interference. A single fluctuation may be caused by noise, but multiple co-occurrences indicate a clear correlation. The co-occurrence determination value can be set as an empirical value. For example, when more than half or 60% of the event pairs meet the synchronization condition in time, it is considered that there is a strong dynamic correlation between the two.

[0094] As another possible approach to the above scheme, the correlation analysis between the plating solution flow rate and the linewidth fluctuation amplitude time lag is as follows: obtain the linewidth fluctuation amplitude sequence and the corresponding plating solution flow rate sequence collected at multiple consecutive time points in the same detection area during the observation period.

[0095] A sliding window is used to perform segmented dynamic analysis of the linewidth fluctuation amplitude sequence and the corresponding plating solution flow rate sequence, and the Pearson correlation coefficient between the flow rate and the linewidth fluctuation amplitude is calculated in each window.

[0096] Considering that the flow rate of the plating solution directly controls the thickness of the diffusion boundary layer on the cathode surface, the higher the flow rate, the thinner the boundary layer, and the faster the mass transfer rate of copper ions to the surface, suppressing concentration polarization. Under this condition, the ion supply at the edge of the line and the groove area is sufficient, suppressing the preferential deposition effect and helping to maintain the consistency of the linewidth. Conversely, when the local flow rate is insufficient, it is easy to cause local ion depletion, triggering non-uniform deposition and causing abnormally large linewidth.

[0097] Therefore, there is a significant correlation between the plating solution flow rate and the linewidth fluctuation range, and a higher flow rate usually corresponds to more stable deposition behavior and smaller linewidth fluctuation, with the two expected to be negatively correlated.

[0098] Because the electroplating process is subject to various disturbances, the relationship between its flow field and quality response is time-varying, and the global correlation may be distorted. The system behavior is approximately stable within a short time window. Therefore, this invention uses a sliding window for correlation analysis. The Pearson correlation coefficient calculated in each window reflects the linear dependence strength between the two within that time period. When the correlation coefficient is negative, it indicates that the flow rate of the plating solution and the linewidth fluctuation amplitude are negatively correlated. The larger the absolute value of the correlation coefficient, the stronger the negative correlation between the variables.

[0099] Within each window, calculate the cross-correlation value between the plating solution flow rate and the linewidth fluctuation amplitude under different time lags, find the lag time offset that makes the cross-correlation value reach the global maximum value, and denote it as the optimal response lag time.

[0100] Given that changes in sediment quality lag behind flow field adjustments, there is a time-delay response between the two. By calculating the cross-correlation function under different time lags and identifying the optimal response lag time, the system inertia from flow field changes to linewidth fluctuation response is reflected. If the optimal response lag time falls within a physically reasonable range, it indicates that the time series relationship conforms to the time scale of the actual physical process, supporting a causal relationship rather than a random correlation between the two.

[0101] Cross-correlation calculations fall within the scope of existing technology and will not be elaborated upon here.

[0102] The linewidth uniformity of the detection area is determined to be strongly correlated with the flow rate of the plating solution when the following conditions are met simultaneously: a) The average Pearson correlation coefficient within the sliding window is negative and the absolute value reaches the correlation threshold.

[0103] The association threshold set above is to exclude weak correlations or random fluctuations, at a given significance level, such as... The critical value of Pearson correlation is calculated based on the sample length. For example, a Pearson correlation coefficient of 0.36 is obtained for 30 data points, which is statistically significant; however, to enhance robustness, a more stringent standard, such as an association threshold of 0.6, is usually adopted.

[0104] b) The optimal response lag time falls within the physically reasonable range.

[0105] The above-mentioned physical reasonable range reflects the inherent dynamic delay between the occurrence of flow field changes and the manifestation of linewidth quality response in the electroplating system, that is, the physical response window of the system. Specifically, after known flow field disturbance events such as pump start-up and shutdown or valve adjustment, the actual delay of linewidth fluctuation response can be observed, an empirical distribution can be constructed, and the 5th to 95th percentile can be taken as the boundary of the reasonable range.

[0106] Understandably, the Pearson correlation coefficient reflects the synergistic trend between linewidth uniformity and plating solution flow rate, while the optimal response lag time characterizes the typical delay in the impact of plating solution flow rate changes on linewidth fluctuations. By setting dual criteria, the reliability of the correlation judgment is significantly improved.

[0107] It should be added that the correlation analysis between current density and copper thickness uniformity, and between plating solution flow rate and linewidth uniformity, uses different technical approaches. The current density and copper thickness uniformity have sudden, step-like disturbances, and the quality degradation has a clear starting point, manifesting as local abrupt changes, which is suitable for event pair co-occurrence analysis. On the other hand, the plating solution flow rate and linewidth fluctuations have continuous, gradual effects, which is suitable for sliding window correlation coefficient and cross-correlation hysteresis analysis.

[0108] In a further feasible manner of the above scheme, the main influencing factors leading to non-uniformity are determined as follows: when only the copper thickness uniformity in the abnormal detection area is strongly correlated with the current density, the current density is determined to be the main influencing factor.

[0109] When the uniformity of line width in the abnormal detection area is strongly correlated with the flow rate of the plating solution, the flow rate of the plating solution is determined to be the main influencing factor.

[0110] When both the abnormal detection area and the plating solution flow rate are strongly correlated, it is determined that the current density and the plating solution flow rate are the main influencing factors.

[0111] Example 2

[0112] See Figure 4 As shown, the present invention proposes a PCB board copper thickness and line width uniformity detection system, including the following modules: Region division module: dividing the PCB board surface into multiple detection areas and establishing a position marker for each detection area.

[0113] Data acquisition module: Connected to the area division module, it uses the electroplating equipment sensors and imaging equipment to acquire the current density, plating solution flow rate and PCB surface image corresponding to each detection area.

[0114] Feature extraction module: Connected to the region segmentation module, it extracts the copper thickness fluctuation amplitude, which represents the uniformity of copper thickness distribution, and the line width fluctuation amplitude, which represents the uniformity of line width size, from the PCB surface image of each detection region.

[0115] Uniformity determination module: Connected to the feature extraction module, it combines spatial proximity comparison and time trend analysis to determine the uniformity of copper thickness fluctuation and linewidth fluctuation in each detection area, and identifies abnormal detection areas with non-uniformity risks.

[0116] Correlation analysis module: Connected to the uniformity determination module and the data acquisition module respectively, for the identified abnormal detection areas, it performs time-series co-occurrence correlation analysis by constructing the current density sequence and the copper thickness fluctuation amplitude sequence, and at the same time, it constructs the plating solution flow rate sequence and the linewidth fluctuation amplitude sequence for time-delay correlation analysis, and further determines the main influencing factors that cause non-uniformity.

[0117] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0118] Those skilled in the art will recognize that the algorithmic steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0119] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0120] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0121] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for detecting the uniformity of copper thickness and line width on a PCB board, characterized in that, Includes the following steps: Step 1: Divide the PCB board into multiple detection areas and establish a location marker for each detection area; Step 2: Use the electroplating equipment's sensors and imaging devices to acquire the current density, plating solution flow rate, and PCB surface image corresponding to each detection area; Step 3: Extract the copper thickness fluctuation amplitude, which represents the uniformity of copper thickness distribution, and the line width fluctuation amplitude, which represents the uniformity of line width size, from the PCB surface image of each detection area; Step 4: Combining spatial proximity comparison and time trend analysis, determine the uniformity of copper thickness fluctuation and linewidth fluctuation in each detection area, and identify abnormal detection areas with non-uniformity risks. Step 5: For the identified abnormal detection areas, a time-series co-occurrence correlation analysis is performed by constructing a current density sequence and a copper thickness fluctuation amplitude sequence. At the same time, a time-delay correlation analysis is performed by constructing a plating solution flow rate sequence and a linewidth fluctuation amplitude sequence, and the main influencing factors causing non-uniformity are further determined. Step 1 is as follows: Based on the topological features of the PCB board's graphic layout, the entire board is functionally partitioned into high graphic density areas and low graphic density areas; within each functional area, several detection regions are generated by dividing the grid into grid cells, wherein the grid division is constrained by the functional area boundaries to ensure that each detection region is completely contained within a single functional area and does not cross the boundaries of different areas; spatial location identifiers are assigned to all generated detection regions, and these location identifiers contain functional area type and grid coordinate information; The copper thickness fluctuation amplitude includes the following extraction process: Based on the PCB surface image of each detection area, the functional area to which it belongs is identified, and the sampling points are arranged according to the functional area as follows: High graphic density area: The midpoint of the upper and lower lines in the detection area is determined and connected to form a central axis. The central axis and the boundary of the detection area together divide the area into four sub-areas; Sampling points are arranged at the geometric center point of the detection area, the center point of the four sub-areas, and the boundary point. Low pattern density area: Sampling points are placed at the geometric center point and the midpoint of the four sides of each detection area; Image grayscale values ​​are extracted at each sampling point in each detection area; the grayscale value of each sampling point is converted into a copper thickness value according to the pre-calibrated correspondence between grayscale value and copper thickness; the range of copper thickness values ​​of all sampling points in each detection area is calculated as the copper thickness fluctuation range of that detection area.

2. The method for detecting the uniformity of copper thickness and line width on a PCB board as described in claim 1, characterized in that: The specific detection process for the current density and plating solution flow rate is as follows: A sensor array is installed inside the electroplating tank to collect current density and plating solution flow rate; Establish a mapping relationship between the sensor acquisition location and the detection area location marker on the PCB board, thereby assigning a corresponding current density and plating solution flow rate to each detection area.

3. The method for detecting the uniformity of copper thickness and line width on a PCB board as described in claim 1, characterized in that: The linewidth fluctuation range includes the following extraction process: Edge detection is performed on the PCB surface image of each detection area to identify the outline of the circuit pattern; Measurement points are set up along the vertical direction of the line contour. At each measurement point, a normal section perpendicular to the tangent direction of the contour is constructed. The line width value at that position is obtained by obtaining the Euclidean distance between corresponding points on both sides. Calculate the range of all linewidth values ​​within each detection area, which is taken as the linewidth fluctuation range of that detection area.

4. The method for detecting the uniformity of copper thickness and line width on a PCB board as described in claim 1, characterized in that: The specific implementation of step 4 is as follows: For each detection area, all detection areas adjacent to its boundary are identified as neighboring areas; Continuously record the copper thickness and linewidth fluctuation amplitudes of each detection area at different time points; Compare the copper thickness fluctuation range and line width fluctuation range of each detection area with the average copper thickness fluctuation range of its neighboring areas; The copper thickness fluctuation range and line width fluctuation range of each detection area are compared with the copper thickness fluctuation range and line width fluctuation range of the previous sampling time. If either the copper thickness fluctuation or the linewidth fluctuation in a certain detection area is higher than the average fluctuation of neighboring areas, or if either the copper thickness fluctuation or the linewidth fluctuation shows an upward trend over multiple consecutive sampling periods, then the detection area is identified as an abnormal detection area.

5. The method for detecting the uniformity of copper thickness and line width on a PCB board as described in claim 1, characterized in that: The time-series co-occurrence correlation analysis performed by constructing a current density sequence and a copper thickness fluctuation amplitude sequence is as follows: Obtain the copper thickness fluctuation amplitude sequence and the corresponding current density sequence of the anomaly detection area collected at multiple consecutive time points during the observation period; The copper thickness fluctuation amplitude sequence and the current density sequence were used to construct copper thickness fluctuation time curves and current density time curves, respectively. Capture the inflection point of the copper thickness fluctuation time curve and extract the rising point of the copper thickness fluctuation. Similarly, inflection point capture is performed on the current density time curve to extract the current density decrease point; The two types of characteristic points, namely the copper thickness fluctuation rise point and the current density decrease point, are time-series aligned and matched on the time axis to construct a set of copper thickness fluctuation rise-current density decrease event pairs. If the time interval between the two types of characteristic points of the copper thickness fluctuation increase point and the current density decrease point in the set falls within the association window, it is determined to be a time-series co-occurrence event. The frequency of time-series co-occurrence events in the anomaly detection area during the observation period is statistically analyzed, and the co-occurrence rate is calculated. If the co-occurrence rate reaches the co-occurrence judgment value, it is determined that there is a strong correlation between the copper thickness uniformity and the current density in the detection area.

6. The method for detecting the uniformity of copper thickness and line width on a PCB board as described in claim 5, characterized in that: The time-delay correlation analysis of the constructed plating solution flow rate sequence and the linewidth fluctuation amplitude sequence is as follows: Obtain the linewidth fluctuation amplitude sequence and the corresponding plating solution flow rate sequence of multiple consecutive time points in the same detection area within the observation period; A sliding window was used to perform segmented dynamic analysis of the linewidth fluctuation amplitude sequence and the corresponding plating solution flow rate sequence, and the Pearson correlation coefficient between the flow rate and the linewidth fluctuation amplitude was calculated in each window. Calculate the cross-correlation value between the plating solution flow rate and the linewidth fluctuation amplitude under different time lags within each window, find the lag time offset that makes the cross-correlation value reach the global maximum value, and record it as the optimal response lag time. The linewidth uniformity of the detection area is considered to be strongly correlated with the plating solution flow rate when the following conditions are met simultaneously: a) The average Pearson correlation coefficient within the sliding window is negative, and the absolute value reaches the correlation threshold; b) The optimal response lag time falls within the physically reasonable range.

7. The method for detecting the uniformity of copper thickness and line width on a PCB board as described in claim 6, characterized in that: The main influencing factors leading to non-uniformity are as follows: When only the copper thickness uniformity and current density are strongly correlated in the abnormal detection area, the current density is determined to be the main influencing factor. When only the line width uniformity of the abnormal detection area is strongly correlated with the plating solution flow rate, the plating solution flow rate is determined to be the main influencing factor. When both the abnormal detection area and the plating solution flow rate are strongly correlated, it is determined that the current density and the plating solution flow rate are the main influencing factors.

8. A PCB board copper thickness and line width uniformity detection system, used to perform the steps of the PCB board copper thickness and line width uniformity detection method according to any one of claims 1-7, characterized in that, Includes the following modules: Region segmentation module: Divides the PCB board surface into multiple detection areas and establishes a location identifier for each detection area; Data acquisition module: Utilizes sensors and imaging equipment from the electroplating equipment to acquire the current density, plating solution flow rate, and PCB surface image corresponding to each detection area; Feature extraction module: Extracts the copper thickness fluctuation amplitude, which characterizes the uniformity of copper thickness distribution, and the linewidth fluctuation amplitude, which characterizes the uniformity of linewidth size, from the PCB surface image of each detection area; Uniformity determination module: Combining spatial proximity comparison and time trend analysis, it determines the uniformity of copper thickness fluctuation and linewidth fluctuation in each detection area, and identifies abnormal detection areas with non-uniformity risks. Correlation analysis module: For the identified abnormal detection areas, time-series co-occurrence correlation analysis is performed by constructing current density sequence and copper thickness fluctuation amplitude sequence. At the same time, time-delay correlation analysis is performed by constructing plating solution flow rate sequence and linewidth fluctuation amplitude sequence, and further the main influencing factors causing non-uniformity are determined.

Images