A method of testing probe performance

Abstract

The present application relates to the technical field of semiconductor probe performance test, especially to a probe performance test method, comprising: through contact simulation test and signal transmission simulation test, calculating the resistance consistency and signal transmission fidelity ratio of probe interference working condition, and carrying out single probe simulation verification; for the double probe composed of qualified single probes, calculating the contact pressure interference quantity, electromagnetic crosstalk interference quantity and carrying out collaborative interference judgment; then extracting historical probe false data, constructing interference false risk model and non-interference false risk model; grouping the double probes according to the interference degree, substituting into the corresponding model to calculate the risk evaluation value; for the high-risk group, determining the adjustment direction and adjustment scheme, calculating the average attenuation rate of the double probes after adjustment; after the attenuation rate verification reaches the standard, it is determined as a qualified performance test probe group, which effectively improves the accuracy, stability and anti-interference ability of semiconductor probe performance detection.

Description

Technical Field

[0001] This invention relates to the field of semiconductor probe performance testing technology, and in particular to a probe performance testing method. Background Technology

[0002] With the deep application of semiconductor manufacturing technology in chip production, wafer inspection and other fields, semiconductor probes, as core components for circuit continuity testing and signal transmission testing, directly determine the accuracy and reliability of the test results. The contact stability and signal transmission fidelity of probes have become key indicators for measuring the practicality of semiconductor testing technology.

[0003] Current semiconductor probe performance testing methods are ill-suited to the demands of complex testing scenarios. For dual-probe applications involving multi-channel synchronous testing, existing methods do not quantify the synergistic interference between the two probes. When the two probes simultaneously contact the wafer, resistance fluctuations can easily occur due to the superposition of contact pressure, or channel crosstalk can be caused by electromagnetic coupling between adjacent probes. These uncontrolled synergistic interferences can increase the detection error. The increased detection error may lead to a higher probability of false positives (such as signal distortion caused by poor contact or misjudgment caused by crosstalk) that may exist in the original single-probe performance testing. This results in insufficient accuracy of probe detection results when performing performance testing with dual probes, which seriously affects the reliability and safety of subsequent semiconductor manufacturing processes. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a probe performance testing method to solve the aforementioned technical problems in the prior art.

[0005] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows: A probe performance testing method, comprising: Through contact simulation tests and signal transmission simulation tests, the resistance consistency and signal transmission fidelity under probe interference conditions are calculated, and the passability of single probe simulation verification is determined. If the judgment is passed, the dual-probe performance test dataset is obtained, dual-probe collaborative interference analysis is performed, the contact pressure interference and electromagnetic crosstalk interference are calculated, and dual-probe collaborative interference judgment is carried out. Based on the dual-probe interference judgment results, target false data are extracted, and a false risk model is constructed. Among them, the false risk model includes: the interference false risk model and the non-interference false risk model; Based on the results of single-probe simulation verification, the risk assessment group of dual probes is determined, the risk assessment group of dual probes is classified, and combined with the false risk model, the risk assessment value of dual probes is output, and the risk assessment analysis of dual probe risk assessment group is carried out. If the risk assessment is high, special determination of single-probe attenuation rate shall be performed to determine the performance test probe group.

[0006] Further, the process of calculating the resistance consistency and signal transmission fidelity ratio under probe interference working conditions is as follows: Based on the contact simulation test, the contact resistance value between the probe and a standard clean wafer, the contact resistance value between the probe and an interference working condition wafer, and the baseline resistance value are extracted; Difference processing is performed on the contact resistance values of the probe with the standard clean wafer and the interference working condition wafer respectively against the baseline resistance value, to obtain a corrected resistance value and a corrected interference resistance value; Ratio processing is performed on the corrected resistance value and the corrected interference resistance value, to obtain the resistance consistency of the probe under interference working conditions; Based on the signal transmission simulation test, the transmission intensity values and blank background signal values of signals at each gradient frequency are extracted; The relative difference between the gradient of transmission intensity values and the gradient of blank background signal values is calculated to obtain the signal transmission fidelity ratio.

[0007] Further, the process of performing pass determination for single-probe simulation verification is as follows: If the resistance consistency is less than the resistance consistency threshold, the contact stability is determined to be qualified; If the signal transmission fidelity ratio is greater than or equal to the signal transmission fidelity ratio threshold, the signal transmission reliability is determined to be qualified; When both the two dimensions of contact stability and signal transmission reliability are determined to be qualified, the single-probe simulation verification is determined to be passed.

[0008] Further, the process of calculating the contact pressure interference amount of the probe is as follows: Based on the dual-probe performance test data set, when the two probes cooperatively contact the wafer, the contact resistance values R_A_co and R_B_co of probe A and probe B, and the baseline resistance value R_blank_co of the dual-probe cooperative system are extracted; The contact resistance values R_A_single and R_B_single of probe A and probe B when they independently contact the wafer, and the baseline resistance value R_blank_single of the single probe are extracted; Baseline correction is performed on R_A_co, R_B_co, R_A_single, and R_B_single, to obtain corrected R_A_co, corrected R_B_co, corrected R_A_single, and corrected R_B_single; Wherein, baseline correction refers to performing difference processing on the probe contact resistance value and the baseline resistance value; The relative difference between corrected R_A_co and corrected R_A_single is calculated to obtain the contact pressure interference amount I_pressure_A of probe A; The relative difference between corrected R_B_co and corrected R_B_single is calculated to obtain the contact pressure interference amount I_pressure_B of probe B.

[0009] Further, the process of calculating the electromagnetic crosstalk interference amount is as follows: Based on the dual-probe performance test dataset, the crosstalk signal value C_A_crosstalk, the crosstalk signal value C_B_crosstalk, and the blank background signal value C_blank_password of the dual-probe cooperative system are extracted. Extract the transmission strength values ​​C_A_single and C_B_single when probe A and probe B work independently, and the blank background signal value C_empty_passage_single when the single probe system is empty; Noise corrections are applied to C_A series coherence, C_B series coherence, C_A single-string coherence, and C_B single-string coherence respectively to obtain C_A series coherence correction, C_B series coherence correction, C_A single-string correction, and C_B single-string correction. The electromagnetic crosstalk interference amount I_stringAB is obtained by comparing the C_A string co-repair with the C_B self-repair. By comparing the C_B string co-repair with the C_A self-repair, the electromagnetic crosstalk interference amount I_stringBA of probe B to probe A is obtained.

[0010] Furthermore, the process of conducting dual-probe collaborative interference decision-making is as follows: Calculate the composition ratio of I_pressureA and I_pressureB respectively to obtain the contact pressure interference coefficient K_pressureAB of probe A to probe B and the contact pressure interference coefficient K_pressureBA of probe B to probe A. Based on signal transmission simulation test, extract the net signal transmission strength values ​​of probe A and probe B; The electromagnetic crosstalk interference coefficient K_stringAB is obtained by comparing the net transmission strength value of the signal from probe A to probe B by I_stringAB and the net transmission strength value of probe A. The electromagnetic crosstalk interference coefficient K_stringBA is obtained by comparing the net transmission strength value of I_stringBA with that of probe B. The mean values ​​of K_pressure AB and K_pressure BA are calculated, and the sum of the mean values ​​of K_string AB and K_string BA is obtained to determine the degree of mutual interference between probe A and probe B. If the degree of mutual interference is greater than or equal to the threshold, it indicates that there is interference in the collaboration of the two probes. Conversely, it indicates that there is no interference or influence.

[0011] Furthermore, the process of constructing a pseudo-risk model is as follows: The target false data include: resistance consistency under probe interference conditions, signal transmission fidelity ratio, and the degree of mutual interference between the two probes; A multiple linear regression model was used to construct both a perturbation false risk model and a non-perturbation false risk model. The target false data is used as the input to the interference false risk model, and the resistance consistency and transmission fidelity after removing the degree of mutual interference are used as the input to the non-interference false risk model. The percentage of false events is calculated by counting the number of false events. The relative deviation between the actual signal and the test signal is calculated and multiplied with the proportion of false events to obtain the false risk value, which is used as the output of the interference false risk model and the non-interference false risk model. Iterative optimization is performed using the least squares method to update the weights of the input model data. The iteration stops when the mean absolute error between the model's predicted false risk value and the actual false risk value is minimized.

[0012] Furthermore, the process of identifying and classifying the dual-probe risk assessment group is as follows: The probes that have passed single-probe simulation verification are combined in pairs to obtain dual-probe risk assessment groups, and the dual-probe risk assessment group set is obtained by summing them up. Extract the degree of mutual interference between the two probes in the dual-probe risk assessment group; The risk assessment group of dual probes with mutual interference levels greater than or equal to the threshold is marked as the dual probe interference risk assessment group. The dual-probe risk assessment group whose mutual interference level is less than the threshold is marked as the dual-probe non-interference risk assessment group.

[0013] Furthermore, combining the false risk model, the process of outputting the dual-probe risk assessment value and conducting risk assessment analysis for the dual-probe risk assessment group is as follows: Based on any dual-probe interference risk assessment group: Extract the resistance consistency, signal transmission fidelity ratio, and mutual interference degree of the two probes in the dual probe interference risk assessment group, and substitute them into the interference false risk model to obtain the dual probe risk assessment value of the dual probe interference risk assessment group. Based on any dual-probe non-interference risk assessment group: Extract the resistance consistency and signal transmission fidelity of the dual probes in the dual probe non-interference risk assessment group, and substitute them into the non-interference false risk model to obtain the dual probe risk assessment value of the dual probe non-interference risk assessment group. If the risk assessment value of the dual probe is greater than or equal to the threshold, it indicates that the risk assessment of the dual probe risk assessment group is high.

[0014] Furthermore, the process of determining the performance test probe group by specifically judging the attenuation rate of a single probe is as follows: The process of determining the performance test probe group by specifically judging the attenuation rate of a single probe is as follows: Perform probe testing and adjustments; Based on any adjusted dual-probe risk assessment group: The relative difference between the contact resistance value of probe A when the two probes work together and the contact resistance value when probe A detects independently is calculated to obtain the attenuation rate of probe A; The relative difference between the contact resistance value of probe B when the two probes work together and the contact resistance value when probe B is used for independent detection is calculated to obtain the attenuation rate of probe B. The average attenuation rate of probe B and probe A is averaged to obtain the average attenuation rate of the two probes. If the average attenuation rate of the dual probes is less than the attenuation qualification threshold, then the dual probes in the dual probe risk assessment group will be used as the performance test probe group.

[0015] The beneficial effects of this invention are: (1) Reduce the false risk of single probe performance testing: Through contact simulation testing and signal transmission simulation testing, the system verifies the specificity and identifiability of single probes in two dimensions, thereby reducing the false risk of single probe performance testing and improving the accuracy of single probe performance testing; (2) Predicting false risks and avoiding potential false risks in test results in advance: By using historical probe false data, construct interference false risk models and non-interference false risk models. By inputting probe test data, the false risk value can be predicted, thus realizing the advance identification of potential probe performance risks. (3) Avoid the false risk of dual probe performance testing caused by dual probe synergistic interference: Through dual probe synergistic interference judgment, risk assessment grouping, targeted adjustment of high-risk group and attenuation rate verification, a complete dual probe performance testing process is formed to obtain the performance test probe group. Performance testing is carried out based on the test data of the performance test probe group to reduce the false risk of dual probe performance testing caused by dual probe synergistic interference. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the probe performance testing method of the present invention; Figure 2 This is a logic diagram of the probe performance testing method of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will be further described in a non-limiting manner below with reference to the accompanying drawings and specific embodiments.

[0018] like Figures 1-2 As shown, this embodiment of a probe performance testing method mainly addresses the cooperative interference between two probes in semiconductor probe testing (such as contact pressure superposition and electromagnetic crosstalk). These interferences may lead to false positives caused by single probes (such as false positives caused by contact resistance fluctuations and false negatives caused by signal transmission distortion), ultimately resulting in insufficient accuracy of probe detection results and seriously affecting the reliability and safety of subsequent semiconductor manufacturing applications. By analyzing the probability of cooperative interference and false positives between two probes, the optimal performance test probe group is selected, specifically including the following steps: Step 1: Calculate the resistance consistency and signal transmission fidelity under probe interference conditions through contact simulation test and signal transmission simulation test, and determine the pass rate of single probe simulation verification. In step one, contact simulation testing refers to a standardized simulation verification process conducted to verify the stable contact capability of the probe under test with a standard wafer and to eliminate the influence of interfering operating conditions (including contaminated wafers, temperature and humidity fluctuations, etc.). Sample preparation: standard clean wafers, wafers with contaminants (such as photoresist residues and metal particles) under interfering conditions, and blank stages (no wafers); After testing the probe and three samples at a preset pressure (preferably 20mN), a resistance tester is used to detect the contact resistance between the probe and the standard clean wafer, the wafer under interference conditions, and the baseline resistance of the blank stage. In step one, the signal transmission simulation test refers to the test process conducted to verify the fidelity of the probe under test when transmitting signals at different frequencies. Sample preparation: Prepare a gradient frequency signal source covering the commonly used frequency range of semiconductor detection (preferably 1kHz-1GHz) and a blank control group with no signal input; Connect the probe to be tested to the signal source according to the preset parameters, collect the transmission intensity value of each gradient frequency signal, repeat the test at each frequency (preferably 3 times) and take the average value to obtain the transmission intensity value of the gradient frequency signal and the blank background signal value of the blank control group. In step one, the process of calculating the resistance consistency under probe interference conditions is as follows: The probe interference conditions refer to non-standard test environments or conditions that affect probe contact stability and signal transmission fidelity in semiconductor probe performance testing, including but not limited to: contact with wafers containing contaminants such as photoresist residue and metal particles, and external environmental interference such as temperature and humidity fluctuations. These conditions can cause fluctuations in probe contact resistance and distortion in signal transmission, which may lead to false test results. Based on contact simulation testing, the contact resistance values ​​between the probe and a standard clean wafer, the contact resistance values ​​between the probe and a wafer under interference conditions, and the baseline resistance value are extracted. First, perform baseline correction on the contact resistance value: The difference between the contact resistance value of the probe and the standard clean wafer and the baseline resistance value is processed to obtain the resistance correction value; The difference between the contact resistance value between the probe and the wafer under interference and the baseline resistance value is processed to obtain the interference resistance correction value. The resistance consistency under probe interference conditions is obtained by comparing the interference resistance correction value with the resistance correction value. It should be noted that the calculation logic for resistance consistency is as follows: based on the three resistance values ​​obtained from the contact simulation test, the influence of equipment error and environmental interference on the real contact signal is first eliminated by subtracting the baseline resistance value. Then, the contact stability of the probe under complex conditions is quantified by the ratio of the resistance value after correction under interference conditions to that under standard conditions, reflecting the probe's specific contact capability with the standard wafer.

[0019] In step one, the process of calculating the signal transmission fidelity is as follows: Based on signal transmission simulation testing, the transmission intensity values ​​and blank background signal values ​​of the probe under test at each gradient frequency are extracted. It should be noted that each gradient concentration needs to be repeatedly detected and the average signal value is taken as the transmission intensity value of the gradient sample after the probe reacts with each gradient sample.

[0020] Based on any gradient frequency: The difference between the transmission strength value and the blank background signal value is processed to obtain the net transmission strength value of the signal. The signal transmission fidelity ratio is obtained by comparing the net transmission strength value with the transmission strength value. In step one, the process of determining the pass rate of single-probe simulation verification is as follows: It should be noted that single-probe simulation verification is not for a single probe, but rather involves independently verifying each probe to be tested. The following is the verification and judgment process for a single probe: Compare the resistance consistency under probe interference conditions with the resistance consistency threshold: If the resistance consistency is less than the resistance consistency threshold, the contact stability is deemed acceptable. If the resistance consistency is greater than or equal to the resistance consistency threshold, the contact stability is deemed unqualified. Based on any gradient frequency: Compare the signal transmission fidelity ratio with a signal transmission fidelity ratio threshold: If the signal transmission fidelity is greater than or equal to the signal transmission fidelity threshold, the signal transmission reliability is deemed acceptable. If the signal transmission fidelity is less than the signal transmission fidelity threshold, the signal transmission reliability is deemed unqualified. The resistance consistency threshold is set according to the relative deviation and error in Part 6 of GB / T 7339-1987 "Fixed Resistors for Electronic Equipment", and the signal transmission fidelity threshold is set according to ITU-T G.122 published by the International Telecommunication Union, which states that "when the effective transmission ratio of the signal is ≥92%, data misjudgment can be avoided".

[0021] When both contact stability and signal transmission reliability are deemed satisfactory, the single-probe simulation verification is considered successful. If any dimension fails to meet the requirements, the single-probe simulation verification is deemed unsuccessful, and the process returns to the probe design optimization stage (such as adjusting the tip material, optimizing the probe elastic coefficient, etc.). After optimization, the full-process verification of step one is re-executed.

[0022] Step 2: If the judgment is passed, obtain the dual-probe performance test dataset, perform dual-probe collaborative interference analysis, calculate the contact pressure interference and electromagnetic crosstalk interference, and make a dual-probe collaborative interference decision. In step two, the dual-probe performance test dataset includes the following data: the contact resistance and baseline resistance of a single probe when the dual probes are working together to contact the wafer, the contact resistance and baseline resistance of a single probe when the probes are contacting independently; the crosstalk signal and blank background signal when the dual probes are working together, and the transmission strength and blank background signal when the probes are working independently. It should be noted that the crosstalk signal value is the cross-channel interference signal generated by the probes through electromagnetic coupling when the two probes work together. It does not belong to the target effective signal of either probe. For example, in a dual-probe test scenario, the transmission strength value of probe A is collected through the detection channel of probe B, or the transmission strength value of probe B is collected through the detection channel of probe A.

[0023] One possible method is to retrieve the data from the semiconductor testing laboratory's data storage center; In step two, the process of calculating the contact pressure interference of the probe is as follows: Based on the dual-probe performance test dataset, the contact resistance value of probe A (denoted as R_A_col) and the contact resistance value of probe B (denoted as R_B_col) and the baseline resistance value of the dual-probe collaborative system (denoted as R_empty_col) are extracted when the dual probes collaboratively contact the wafer. Based on the dual-probe performance test dataset, the contact resistance value when probe A independently contacts the wafer (denoted as R_A_single), the contact resistance value when probe B independently contacts the wafer (denoted as R_B_single), and the baseline resistance value of a single probe (denoted as R_empty_single) are extracted. The difference between R_A and R_empty is processed to obtain the cooperative resistance correction value R_A of probe A; The difference between R_B_coordinate and R_empty_coordinate is processed to obtain the cooperative resistance correction value R_B_coordinate of probe B. The difference between R_A single and R_ empty single is processed to obtain the resistance correction value R_A single correction of probe A; The difference between R_B single and R_ empty single is processed to obtain the resistance correction value R_B single correction of probe B; Calculate the absolute value of the difference between R_A collaborative repair and R_A single repair, and perform ratio processing with R_A single repair to obtain the contact pressure interference amount I_pressureA of probe A; Calculate the absolute value of the difference between R_B collaborative repair and R_B single repair, and compare it with R_B single repair to obtain the contact pressure interference amount I_pressureB of probe B (the larger the value, the more significant the interference of the probe is caused by the binding competition of another probe). The process of calculating electromagnetic crosstalk is as follows: Based on the dual-probe performance test dataset, the crosstalk signal value of probe A (the transmission strength value of probe A collected in the detection channel of probe B) is extracted and denoted as C_A crosstalk. The crosstalk signal value of probe B in the detection channel of probe A (the transmission strength value of probe B collected in the detection channel of probe A) is denoted as C_B crosstalk. The blank background signal value of the dual-probe cooperative system is denoted as C_space-pass cooperative; It should be noted that a single-probe system refers to the general term for testing with a single probe, including testing with single probes A and B, while a dual-probe system refers to testing with two probes.

[0024] The transmission strength value of probe A when it works independently is denoted as C_A_single. The transmission strength value of probe B when it works independently is denoted as C_B_single. The blank background signal value of a single probe system is denoted as C_empty single; The difference between C_A string coherence and C_space coherence is processed to obtain C_A string coherence; The difference between the C_B string coherence and the C_air passage coherence is processed to obtain the C_B string coherence repair; The difference between C_A self-order and C_air-pass order is processed to obtain C_A self-order repair; The difference between C_B self-order and C_air-pass order is processed to obtain C_B self-order repair; The electromagnetic crosstalk interference amount I_stringAB is obtained by comparing the C_A string co-repair with the C_B self-repair. The electromagnetic crosstalk interference amount I_stringBA is obtained by comparing the C_B string co-repair with the C_A self-repair. In step two, the process of conducting dual-probe collaborative interference decision-making is as follows: Add I_pressure A and I_pressure B to obtain the total contact pressure interference of the probe; The contact pressure interference coefficient K_pressureAB of probe A to probe B is obtained by comparing I_pressureA with the total contact pressure interference. The contact pressure interference coefficient K_pressureBA of probe B to probe A is obtained by comparing I_pressureB with the total contact pressure interference. Based on signal transmission simulation test, extract the net signal transmission strength values ​​of probe A and probe B; The electromagnetic crosstalk interference coefficient K_stringAB is obtained by comparing the net transmission strength value of the signal from probe A to probe B by I_stringAB and the net transmission strength value of probe A. The electromagnetic crosstalk interference coefficient K_stringBA is obtained by comparing the net transmission strength value of I_stringBA with that of probe B. The mean values ​​of K_pressure AB and K_pressure BA are averaged and then added to the mean values ​​of K_string AB and K_string BA to obtain the degree of mutual interference between probe A and probe B. Compare the mutual interference level of probe A and probe B with the mutual interference level threshold: The threshold for mutual interference is set according to GB / T 7339-1987 "Fixed Resistors for Electronic Equipment" which states that "when multiple units of a resistor network work together, the performance fluctuation caused by mutual interference is ≤10%". If the degree of mutual interference is greater than or equal to the threshold of mutual interference, it indicates that there is interference in the collaboration of the two probes. If the degree of mutual interference is less than the mutual interference threshold, it indicates that there is no interference from the dual probe collaboration.

[0025] The beneficial effects of conducting dual-probe coordinated interference decision-making are: Beneficial effect 1: By quantifying the amount of contact pressure interference, electromagnetic crosstalk interference, and corresponding interference coefficients, the type and intensity of interference when the dual probes work together can be accurately identified. Beneficial effect 2: As a distinguishing condition for the subsequent construction of interference false risk models and non-interference false risk models, by judging whether there is interference when the two probes work together, the single probe false dataset and the two probe false dataset can be distinguished and used as the model inputs for interference false risk models and non-interference false risk models, respectively. Benefit 3: As the main basis for subsequently determining the probe test adjustment targets, by using the dual-probe collaborative interference decision conditions, it can distinguish between single-probe false datasets and dual-probe false datasets, and select different corresponding probe test adjustment targets accordingly.

[0026] Step 3: Based on historical probe test results, extract target false data and combine it with the dual-probe interference judgment results to construct a false risk model; Among them, the false risk model includes: the interference false risk model and the non-interference false risk model; In step three, the target false data refers to: false events that deviate from the true detection results in single-probe and dual-probe tests: probe test data when a false event occurs. False events include false positive events (detection of a conduction signal when there is no target detection point) and false negative events (failure to detect a conduction signal when there is a target detection point). The test data includes: resistance consistency under probe interference conditions, signal transmission fidelity ratio, and mutual interference between the two probes. Obtain probe test data for false events that occurred within the historical statistical period, and categorize them according to the number of probes: The historical statistics duration refers to the time period from the start of the current test scenario to the current point in time. When a false event occurs with a single probe, the corresponding probe test data is marked as false data for a single probe. When a false event occurs with the dual probes, the corresponding probe test data is marked as false dual probe data. The probe test data that showed false events were summarized according to the same category to obtain single probe false datasets and double probe false datasets; Single-probe false datasets and dual-probe false datasets are collectively referred to as target false datasets; In step three, the process of constructing the pseudo-risk model is as follows: The interference false risk model refers to the model that takes the resistance consistency, transmission fidelity ratio and mutual interference degree of the dual probe interference condition as input, and outputs the false risk value of the dual probe test. Based on any pair of dual probes in the dual probe pseudo dataset, extract the single probe pseudo data corresponding to the dual probes from the single probe pseudo dataset; Based on the false data of a single probe, the resistance consistency and signal transmission fidelity under probe interference conditions are extracted. Based on the false data from the dual probes, the degree of mutual interference is extracted; The resistance consistency, dual signal transmission fidelity, and mutual interference between the two probes under the dual-probe interference condition are recorded as a set of model inputs. The calculation process for the false risk value is as follows: Based on the dual-probe false data set, the proportion of false events (P, i.e., the number of false events / the total number of tests) and the degree of false events (the ratio of the deviation between the actual signal and the test signal to the actual signal) are extracted. The false event risk value is obtained by multiplying the proportion of false events by the severity of false events. A multiple linear regression model was used to construct a model for the false risk of interference. The matching dataset of input variables and output variables is divided into training set and validation set in a 7:3 ratio; The training set data is iteratively optimized using the least squares method to solve the weights of each input parameter in the interference false risk model. The iteration stops when the mean absolute error between the false risk value predicted by the model and the actual false risk value in the training set is minimized. It should be noted that Pearson correlation coefficient analysis can verify the correlation logic between input and output variables: resistance consistency is positively correlated with false risk value (the higher the resistance consistency, the greater the impact of interference conditions on contact stability, and the higher the false risk); signal transmission fidelity is negatively correlated with false risk value (the higher the signal transmission fidelity, the lower the probability of signal distortion, and the lower the false risk); mutual interference degree is positively correlated with false risk value (the stronger the interference, the higher the false risk).

[0027] The process of constructing the non-intrusive pseudo-risk model is as follows: The resistance consistency and dual signal transmission fidelity under dual probe interference conditions are recorded as a set of model inputs; A non-intrusive false risk model was constructed using a multiple linear regression model. The matching dataset of input variables and output variables is divided into training set and validation set in a 7:3 ratio; The training set data is iteratively optimized using the least squares method to solve for the weights of each input parameter in the non-interference false risk model, thereby minimizing the mean absolute error between the false risk value predicted by the model and the actual false risk value in the training set.

[0028] Step 4: Based on the single-probe simulation verification results, determine the dual-probe risk assessment group, classify the dual-probe risk assessment group, and combine it with the false risk model to output the dual-probe risk assessment value and conduct risk assessment analysis of the dual-probe risk assessment group. In step four, the process of determining the dual-probe risk assessment set is as follows: The probes that have passed the single-probe simulation verification are combined in pairs to obtain the dual-probe risk assessment group; The dual-probe risk assessment groups are summarized to obtain the dual-probe risk assessment group set; In step four, the process of classifying the dual-probe risk assessment group is as follows: Extract the degree of mutual interference in the dual-probe risk assessment group; The risk assessment group of dual probes whose mutual interference level is greater than or equal to the mutual interference level threshold is marked as the dual probe interference risk assessment group. The dual-probe risk assessment group whose mutual interference level is less than the mutual interference level threshold is marked as the dual-probe non-interference risk assessment group. In step four, the process of outputting the dual-probe risk assessment value by combining the false risk model is as follows: Based on any dual-probe interference risk assessment group: Extract the resistance consistency, signal transmission fidelity ratio, and mutual interference degree of the two probes in the dual probe interference risk assessment group, and substitute them into the interference false risk model to obtain the dual probe risk assessment value of the dual probe interference risk assessment group. Based on any dual-probe non-interference risk assessment group: Extract the resistance consistency and signal transmission fidelity of the dual probes in the dual probe non-interference risk assessment group, and substitute them into the non-interference false risk model to obtain the dual probe risk assessment value of the dual probe non-interference risk assessment group. In step four, the risk assessment analysis process for the dual-probe risk assessment group is as follows: The risk assessment values ​​of the dual-probe interference risk assessment group and the dual-probe non-interference risk assessment group were compared with the dual-probe risk assessment threshold, respectively: The risk assessment threshold for dual probes is set according to the risk control requirements of Clause 6.2 of GB / T 4589.1-2006; If the risk assessment value of the dual probe is greater than or equal to the risk assessment threshold of the dual probe, it indicates that the risk assessment of the dual probe risk assessment group is high. If the risk assessment value of the dual probe is less than the risk assessment threshold of the dual probe, it means that the risk assessment of the dual probe risk assessment group is low. The dual probe risk assessment group will be subject to a special judgment on the single probe attenuation rate. Those that meet the standard can be directly used as the performance test probe group, and those that do not meet the standard will be eliminated.

[0029] Step 5: If the risk assessment is high, based on the dual-probe collaborative interference judgment result and the single-probe simulation verification pass judgment result, determine the probe test adjustment object and adjustment method, and determine the performance test probe group through the single probe attenuation rate special judgment; In step five, the process of determining the probe test adjustment object and adjustment method is as follows: Extract the dual-probe risk assessment group whose dual-probe risk assessment value is greater than or equal to the dual-probe risk assessment threshold; If the dual-probe risk assessment group belongs to the dual-probe interference risk assessment group: Extract the mean values ​​of K_pressure AB and K_pressure BA, and the mean values ​​of K_string AB and K_string BA, and compare them: If the average values ​​of K_pressure AB and K_pressure BA are greater than or equal to the average values ​​of K_string AB and K_string BA, then the probe test adjustment object is the contact pressure interference. The adjustment method can be: adjust the probe contact pressure ratio (e.g., the original probe A:B pressure ratio is 1:1, which can be adjusted to 2:1), and the adjustment amount is the difference between the mutual interference degree threshold and the average values ​​of K_string AB and K_string BA. If the average values ​​of K_pressure AB and K_pressure BA are less than the average values ​​of K_string AB and K_string BA, then the probe test adjustment target is electromagnetic crosstalk interference. The adjustment method can be: optimizing the probe spacing or using shielding materials (such as plating a shielding layer on the probe surface). If the dual-probe risk assessment group belongs to the dual-probe non-interference risk assessment group: Extract the single-probe simulation verification pass rate determination results of the dual-probe system; Based on any probe: The difference between the resistance consistency threshold and the resistance consistency under probe interference conditions is processed to obtain the probe resistance consistency difference. The difference between the signal transmission fidelity ratio and the signal transmission fidelity ratio threshold is processed to obtain the signal transmission fidelity ratio difference. The single probe contact quality is obtained by comparing the probe resistance consistency difference with the resistance consistency threshold. The single probe transmission quality is obtained by comparing the difference in signal transmission fidelity ratio with the signal transmission fidelity ratio threshold. The sum of the single-probe contact goodness of the dual probes is calculated to obtain the dual-probe contact goodness. The sum of the single-probe transmission superiority of the dual probes is calculated to obtain the dual-probe transmission superiority. If the contact quality of the dual probes is greater than or equal to the transmission quality of the dual probes, then the probe test adjustment object is the resistance consistency of the probes. The adjustment method can be: optimizing the tip shape (such as polishing the tip flatness). If the contact quality of the dual probes is less than the transmission quality of the dual probes, then the probe test adjustment object is the signal transmission fidelity of the probes. The adjustment method can be: replacing the probes with highly conductive materials or optimizing the signal transmission lines. In step five, the process of determining the performance test probe set is as follows: Based on any adjusted dual-probe risk assessment group: Calculate the attenuation rate (R_A) of probe A: Calculate the attenuation ratio of the contact resistance value of probe A when the two probes work together to the contact resistance value when probe A is used for independent detection (the ratio of the difference in contact resistance value to the contact resistance value when probe A is used for independent detection). Probe B attenuation rate (R_B): Calculate the attenuation ratio of the contact resistance value of probe B when the two probes work together to the contact resistance value when probe B detects independently; The average attenuation rate of the dual probes is obtained by averaging R_A and R_B. The average attenuation rate of the dual probes in the dual probe risk assessment group was compared with the acceptable attenuation threshold: The attenuation threshold is set according to the "acceptable range of contact resistance attenuation that has no significant impact on signal transmission of electronic equipment" in GB / T 17626 "Electromagnetic Compatibility Testing and Measurement Technology" Part 9 Pulse Magnetic Field Immunity Test and the "performance attenuation should be controlled within 5% to avoid signal distortion" in IEC62132-4 "Electromagnetic Immunity Test for Semiconductor Devices" Part 4 Direct RF Power Injection Method. If the average attenuation rate of the dual probes is greater than or equal to the attenuation qualification threshold, then the dual probes in the dual probe risk assessment group will be optimized. If the average attenuation rate of the dual probes is less than the attenuation qualification threshold, then the dual probes in the dual probe risk assessment group will be used as the performance test probe group. The working principle of this invention is as follows: First, a single probe is used for initial screening. Contact simulation tests (calculating resistance consistency) and signal transmission simulation tests (calculating signal transmission fidelity) are conducted to determine whether a single probe is qualified. If it is not qualified, it is returned for optimization. For dual probes composed of qualified single probes, the contact pressure interference and electromagnetic crosstalk interference are quantified based on the test dataset. Collaborative interference judgment is carried out through indicators such as interference coefficient and total interference to determine whether the interference is acceptable. Then, historical false positive and false negative data of single and dual probes are extracted, and corresponding risk models are constructed to output false risk values. Subsequently, dual probe groups are classified according to the degree of interference, and the corresponding risk models are substituted to calculate the evaluation value. The value is compared with the threshold to determine the level of risk. For high-risk groups, the adjustment objects are determined according to the type of interference. After optimization, the average attenuation rate of dual probes is calculated. Those with the attenuation rate meeting the standard are the performance test probe groups.

[0030] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and does not limit the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should fall within the scope of the present invention.

Claims

1. A probe performance testing method, characterized in that, include: Through contact simulation tests and signal transmission simulation tests, the resistance consistency and signal transmission fidelity under probe interference conditions are calculated, and the passability of single probe simulation verification is determined. If the judgment is passed, the dual probe performance test dataset is obtained, dual probe cooperative interference analysis is performed, the contact pressure interference and electromagnetic crosstalk interference are calculated, and dual probe cooperative interference judgment is carried out. Based on historical probe test results, target false data are extracted, and combined with the dual-probe interference judgment results, a false risk model is constructed. Among them, the false risk model includes: the interference false risk model and the non-interference false risk model; Based on the results of single-probe simulation verification, the risk assessment group of dual probes is determined, the risk assessment group of dual probes is classified, and combined with the false risk model, the risk assessment value of dual probes is output, and the risk assessment analysis of dual probe risk assessment group is carried out. If the risk assessment value of the dual probe is greater than or equal to the threshold, it indicates that the risk assessment of the dual probe risk assessment group is high. If the risk assessment is high, a specific determination of the single-probe attenuation rate is conducted to identify the performance test probe group. The process is as follows: Perform probe testing and adjustments; Based on any adjusted dual-probe risk assessment group: The relative difference between the contact resistance value of probe A when the two probes work together and the contact resistance value when probe A detects independently is calculated to obtain the attenuation rate of probe A; The relative difference between the contact resistance value of probe B when the two probes work together and the contact resistance value when probe B is used for independent detection is calculated to obtain the attenuation rate of probe B. The attenuation rate of probe B and the attenuation rate of probe A are averaged to obtain the average attenuation rate of the two probes. If the average attenuation rate of the dual probes is less than the attenuation qualification threshold, then the dual probes in the dual probe risk assessment group will be used as the performance test probe group. The process of constructing a pseudo-risk model is as follows: The target false data include: resistance consistency under probe interference conditions, signal transmission fidelity ratio, and the degree of mutual interference between the two probes; A multiple linear regression model was used to construct both a perturbation false risk model and a non-perturbation false risk model. The target false data is used as the input to the interference false risk model, and the resistance consistency and transmission fidelity after removing the degree of mutual interference are used as the input to the non-interference false risk model. The percentage of false events is calculated by counting the number of false events. False events include false positive events where a conduction signal is detected when there is no target detection point and false negative events where a conduction signal is not detected when there is a target detection point. The relative deviation between the actual signal and the test signal is calculated and multiplied with the proportion of false events to obtain the false risk value, which is used as the output of the interference false risk model and the non-interference false risk model. The weights of the input perturbation false risk model data are updated through iterative optimization using the least squares method. The iteration stops when the mean absolute error between the false risk value predicted by the perturbation false risk model and the actual false risk value is minimized. The training set data is iteratively optimized using the least squares method to solve for the weights of each input parameter in the non-interference false risk model, thereby minimizing the mean absolute error between the false risk value predicted by the model and the actual false risk value in the training set.

2. The probe performance testing method according to claim 1, characterized in that, The process of calculating the resistance consistency and signal transmission fidelity under probe interference conditions is as follows: Based on the contact simulation test, extract the contact resistance value between the probe and a standard clean wafer, the contact resistance value between the probe and a wafer under interference conditions, and the baseline resistance value; Perform difference processing on the contact resistance values of the probe with the standard clean wafer and the wafer under interference conditions respectively with the baseline resistance value, to obtain a corrected resistance value and a corrected interference resistance value; Perform ratio processing on the corrected resistance value and the corrected interference resistance value to obtain the resistance consistency of the probe under interference conditions; Based on the signal transmission simulation test, extract the transmission intensity values of each gradient frequency signal and the blank background signal value; Calculate the relative difference between the transmission intensity value and the gradient of the blank background signal value to obtain the signal transmission fidelity ratio.

3. The probe performance testing method according to claim 1, characterized in that, The process for performing pass judgment on single-probe simulation verification is: If the resistance consistency is less than the resistance consistency threshold, the contact stability judgment is determined to be qualified; If the signal transmission fidelity ratio is greater than or equal to the signal transmission fidelity ratio threshold, the signal transmission reliability judgment is determined to be qualified; When both the two dimensions of contact stability and signal transmission reliability are determined to be qualified, the single-probe simulation verification is determined to be passed.

4. The probe performance testing method according to claim 1, characterized in that, The process for calculating the contact pressure interference amount of the probe is: Based on the dual-probe performance test data set, when two probes cooperate to contact the wafer, extract the contact resistance values R_A-co and R_B-co of probe A and probe B, and the baseline resistance value R_blank-co of the dual-probe cooperative system; Extract the contact resistance values R_A-sin and R_B-sin when probe A and probe B independently contact the wafer, and the baseline resistance value R_blank-sin of the single probe; Perform baseline correction on R_A-co, R_B-co, R_A-sin, and R_B-sin to obtain R_A-co-cor, R_B-co-cor, R_A-sin-cor, and R_B-sin-cor; Wherein, baseline correction refers to: performing difference processing on the probe contact resistance value and the baseline resistance value; Calculate the relative difference between R_A-co-cor and R_A-sin-cor to obtain the contact pressure interference amount I_press-A of probe A; Calculate the relative difference between R_B-co-cor and R_B-sin-cor to obtain the contact pressure interference amount I_press-B of probe B.

5. The probe performance testing method according to claim 4, characterized in that, The process for calculating the electromagnetic crosstalk interference amount is: Based on the dual-probe performance test data set, extract the crosstalk signal value C_cros-co of probe A, the crosstalk signal value C_cros-co of probe B, and the blank background signal value C_blank-trans-co of the dual-probe cooperative system; Extract the transmission intensity values C_self-sin-A of probe A and C_self-sin-B of probe B when they work independently, and the blank background signal value C_blank-trans-sin of the single-probe system; Perform noise correction on C_A-cros-co, C_B-cros-co, C_A-self-sin, and C_B-self-sin respectively to obtain C_A-cros-co-cor, C_B-cros-co-cor, C_A-self-sin-cor, and C_B-self-sin-cor; Perform ratio processing on C_A-cros-co-cor and C_B-self-sin-cor to obtain the electromagnetic crosstalk interference amount I_cros-AB of probe A to probe B; Perform ratio processing on C_B-cros-co-cor and C_A-self-sin-cor to obtain the electromagnetic crosstalk interference amount I_cros-BA of probe B to probe A.

6. The probe performance testing method according to claim 1, characterized in that, The process for carrying out dual-probe cooperative interference judgment is: Calculate the composition ratio of I_press-A and I_press-B respectively to obtain the contact pressure interference coefficient K_press-AB of probe A to probe B and the contact pressure interference coefficient K_press-BA of probe B to probe A; Based on signal transmission simulation test, extract the net signal transmission strength values ​​of probe A and probe B; The electromagnetic crosstalk interference coefficient K_stringAB is obtained by comparing the net transmission strength value of the signal from probe A to probe B by I_stringAB and the net transmission strength value of probe A. The electromagnetic crosstalk interference coefficient K_stringBA is obtained by comparing the net transmission strength value of I_stringBA with that of probe B. The mean values ​​of K_pressure AB and K_pressure BA are calculated, and the sum of the mean values ​​of K_string AB and K_string BA is obtained to determine the degree of mutual interference between probe A and probe B. If the degree of mutual interference is greater than or equal to the threshold, it indicates that there is interference in the collaboration of the two probes. Conversely, it indicates that there is no interference or influence.

7. The probe performance testing method according to claim 1, characterized in that, The process of identifying and classifying the dual-probe risk assessment group is as follows: The probes that have passed single-probe simulation verification are combined in pairs to obtain dual-probe risk assessment groups, and the dual-probe risk assessment group set is obtained by summing them up. Extract the degree of mutual interference between the two probes in the dual-probe risk assessment group; The risk assessment group of dual probes with mutual interference levels greater than or equal to the threshold is marked as the dual probe interference risk assessment group. The dual-probe risk assessment group whose mutual interference level is less than the threshold is marked as the dual-probe non-interference risk assessment group.

8. The probe performance testing method according to claim 7, characterized in that: The process of combining the false risk model, outputting the dual-probe risk assessment value, and conducting risk assessment analysis for the dual-probe risk assessment group is as follows: Based on any dual-probe interference risk assessment group: Extract the resistance consistency, signal transmission fidelity ratio, and mutual interference degree of the two probes in the dual probe interference risk assessment group, and substitute them into the interference false risk model to obtain the dual probe risk assessment value of the dual probe interference risk assessment group. Based on any dual-probe non-interference risk assessment group: The resistance consistency and signal transmission fidelity of the dual probes in the dual probe non-interference risk assessment group are extracted and substituted into the non-interference false risk model to obtain the dual probe risk assessment value of the dual probe non-interference risk assessment group.

Images