Silicon carbide epitaxial wafer mercury probe C-V test method based on defect pre-detection

Abstract

The invention discloses a silicon carbide epitaxial wafer mercury probe C-V test method based on defect pre-detection, and the method comprises the steps: firstly carrying out the cleaning treatment of a to-be-tested sample, then carrying out the specific region or global scanning recognition of defects through a defect detector, and excluding initial abnormal test points located in a defect region or closer to the defect region; then redefining a replacement point location in a defect-free area around the abnormal point location, and synchronizing the replacement point location to a mercury probe test system; and after the electrical parameter test is completed according to the replacement point location, repeating the process for the data deviating from the normal range until effective data is obtained. Defect pre-detection and test point location optimization are deeply combined, and a'defect-point location-efficiency-risk 'multi-angle cooperative control mechanism is established; the method does not need large-scale modification of existing hardware, can be integrated in an existing test system, greatly reduces the number of repeated tests, reduces the mercury leakage risk, prolongs the service life of related spare parts, improves the accuracy of test data, is suitable for industrial improvement, and has economic popularization.

Description

Technical Field

[0001] This invention relates to the field of semiconductor material electrical parameter testing technology, and in particular to a mercury probe CV testing method for silicon carbide epitaxial wafers based on defect pre-detection. Background Technology

[0002] Mercury probe chromatographic (CV) testing is a core method for characterizing the electrical parameters of semiconductor materials (such as carrier concentration, doping distribution, and interface state density). Due to its advantages, including no need for metal evaporation, high testing efficiency, and the ability to perform in-situ non-destructive testing, it is widely used in the research, development, production, and quality control of semiconductor materials and devices. Its core principle involves forming a Schottky contact between a mercury probe and the sample surface, applying a bias voltage, and measuring the capacitance change to obtain the material's electrical parameters.

[0003] However, existing mercury probe CV testing technology has the following key problems: Abnormal points lead to repeated testing and low efficiency: In traditional testing procedures, if the test point falls in a defect area, it will cause abnormal test data, requiring the selection of a new point for repeated testing. This prolongs the testing time for a single sample, severely reducing testing efficiency. High risk of mercury leakage and poor safety: The risk of mercury leakage from the mercury probe is positively correlated with the number of probe contacts. During repeated testing, the mercury probe needs to frequently contact the sample surface, which not only increases the risk of mercury dripping and leakage, but may also damage the probe tip due to collisions with defect areas. Repeated adjustments to contact pressure and position further increase the risk of mercury leakage, posing a threat to operator health and the environment. Key spare parts wear out quickly and have a short lifespan: During frequent contact and adjustment, the capillary and metal wire of the mercury probe are prone to wear due to friction and impact, significantly increasing testing costs; Existing technologies have not solved the problem of "eliminating anomalies at the source": Current related improvement technologies mostly focus on "probe structure optimization" or "post-test data correction", but none of them solve the problem at the source of "pre-test defect detection and avoidance of abnormal point selection", and cannot fundamentally improve testing efficiency and reduce the risk of mercury leakage.

[0004] Chinese patent CN202411290629.4 discloses a non-destructive mercury probe testing device. This patent modifies the testing structure on the basis of a traditional structure to achieve the same testing effect. However, this patent relies on hardware modification, the lifespan of the device's spare parts is unknown, and its widespread adoption is limited.

[0005] Chinese patent CN202311131858.7 discloses a method, apparatus, electronic device, and storage medium for correcting doping concentration. This patent improves data accuracy by correcting test data. However, this method cannot filter out actual outliers, thus failing to effectively remove outliers and supplement valid data points for further data processing, resulting in insufficient testing efficiency.

[0006] Japanese Patent JP2016238839 discloses a CV measurement method that can more reliably determine the resistivity of semiconductor single-crystal wafers. However, this method is only a formula derivation under ideal conditions and does not provide further explanation regarding defects, locations, testing efficiency, risks, etc., under actual sample testing conditions. Summary of the Invention

[0007] This application addresses the core pain points of existing mercury probe CV testing for silicon carbide epitaxial wafers, such as numerous anomalies due to lack of pre-detection of defects, low efficiency of repeated testing, increased risk of mercury leakage due to frequent probe contact, rapid wear and short lifespan of spare parts, and the failure of existing technologies to solve the problem at its source. It provides a mercury probe CV testing method for silicon carbide epitaxial wafers based on defect pre-detection. By combining a defect detector with optimized test points, it aims to eliminate anomalies at the source, reduce repeated operations, reduce the risk of mercury leakage, extend the lifespan of spare parts, and reduce costs.

[0008] A mercury probe CV testing method for silicon carbide epitaxial wafers based on defect pre-detection, comprising the following steps:

[0009] Step 1: Prepare the test sample and clean the sample surface for subsequent surface defect testing and mercury probe CV testing;

[0010] Step 2: Use an optical defect detector to perform a full-area scan of the surface of the test sample, identify the number and coordinates of defects, and mark abnormal points;

[0011] Step 3: For the abnormal points excluded in Step 2, automatically exclude the points, do not perform mercury probe contact tests, and redefine alternative points;

[0012] Step 4: Write the alternative point coordinates from Step 3 into the mercury probe testing system, and fix the test sample on the stage of the testing system to perform mercury probe CV testing.

[0013] Step 5: Perform a second round of anomaly detection and verification on the points from Step 4. If the point is still determined to be an anomaly, repeat Steps 2 to 4 until valid data is obtained. After all points have been tested, continue with the above process for the next test sample.

[0014] As a preferred technical solution of this application: the test sample selected in step one can be a semiconductor substrate, a semiconductor epitaxial wafer, or a device chip, and the test object is usually a silicon carbide epitaxial wafer material. Ultrasonic cleaning can be performed using solutions such as acetone, isopropanol, and hydrogen peroxide. The cleaning time can be controlled within 10 minutes, the power can be controlled below 100W, and the flow rate of nitrogen blowing can be controlled at 5L / min to ensure that the surface is free of contamination and does not damage the surface of the epitaxial wafer.

[0015] As a preferred technical solution of this application: in step two, the scanning resolution of the optical defect detector is not less than 1μm, the whole area is marked as a 2x2 mm2 matrix cell, the adjacent spacing is not less than 10mm, to ensure the defect detection accuracy and the rationality of the initial point distribution, for feature defects with a length and width greater than 5 μm, the defect grid is defined according to its maximum coverage range, and the edge range can be set to 3mm after deducting the edge range;

[0016] As a preferred technical solution of this application: in step three, the replacement point should be selected from the four defect-free matrix cells closest to the original abnormal point. The distance between the replacement point and the original abnormal point should be controlled at 2mm, and the distance between all replacement points should not be less than twice the diameter of the mercury probe to avoid mutual interference between test points. After the defect detector confirms for the second time that there are no defects in the area where the replacement point is located, the coordinates of the replacement point are synchronized to the control module of the mercury probe CV test system.

[0017] As a preferred technical solution of this application: in step four, the test environment temperature is controlled at 22±2℃ and the relative humidity is < 50% RH. During the test, the test frequency, contact pressure and contact time are kept constant to ensure uniform test conditions, reduce the influence of environmental and operational factors on the data, collect the capacitance-voltage (CV) characteristic data of the sample, and automatically store the test data and corresponding point coordinates.

[0018] As a preferred technical solution of this application: the abnormal point determination in step five must simultaneously meet three conditions: "the capacitance curve does not conform to the theoretical trend", "the breakdown voltage is not within the normal range" and "the data deviates from the average value by ≥10%", and the abnormal point locations must be re-screened and tested in steps two to four to ensure the validity of the final data.

[0019] As a preferred technical solution of this application: when retesting the original abnormal point in step five, the selection criteria for the replacement point, the coordinate synchronization method, and the secondary defect confirmation process must be completely consistent with the requirements for the initial selection of the replacement point to ensure the consistency between the retested data and the normal test data.

[0020] Explanation of the principle:

[0021] By using defect detection equipment, the entire testing area can be identified for defects, and the coordinates of the defects at the testing points can be defined. This eliminates the risk of abnormal point contact at the source and avoids the inefficient process of "testing first, then finding the abnormality, and then repeating" in traditional methods. "Dynamic point redefinition" ensures that there are no missing parts in the testing area and guarantees data accuracy. By controlling the contact pressure, time, and ambient temperature and humidity of the mercury probe, the number of invalid contacts between the probe and the sample is reduced, and the wear and tear on spare parts caused by invalid operations is reduced. Finally, a closed-loop process of "detection-optimization-testing-protection" is constructed to solve multiple pain points of existing technologies. The testing method can be integrated with the original mercury probe system through software modules without the need for hardware modification, thereby improving the level of automation.

[0022] The mercury probe CV testing method for silicon carbide epitaxial wafers based on defect pre-detection described in this application has the following technical advantages compared with the prior art:

[0023] This solution boasts multiple core advantages. For the first time, it deeply integrates "defect pre-detection" with "mercury probe CV test point optimization," establishing a collaborative control mechanism encompassing "defects—test points—test efficiency—risk," filling a gap in existing technologies for "source anomaly elimination." It avoids the time wasted on ineffective retests in traditional testing and automatically defines defect points without manual intervention, significantly improving overall testing efficiency and adapting to high-volume sample testing scenarios. Eliminating defect points "before contact" prevents mercury droplets from contacting uneven surfaces, reducing the risk of mercury leakage and minimizing pollution of the testing environment. It also avoids collisions between the probe and defect areas, enhancing operational safety. Furthermore, it reduces wear on critical spare parts, significantly extending consumable lifespan and lowering testing costs. Through defect elimination rules and spacing control at alternative test points, the accuracy and reliability of test data are significantly improved, meeting the high-precision requirements of semiconductor mass production quality control. Attached Figure Description

[0024] Figure 1 This is a schematic flowchart of the mercury probe CV test method for silicon carbide epitaxial wafers based on defect pre-detection, as described in this application.

[0025] Figure 2 This is a schematic diagram of the test apparatus for the mercury probe CV test method for silicon carbide epitaxial wafers based on defect pre-detection, as described in this application. Detailed Implementation

[0026] The technical solution of the present invention will be described in detail below, but the scope of protection of the present invention includes, but is not limited to, the embodiments described.

[0027] Example 1:

[0028] A mercury probe CV testing method for silicon carbide epitaxial wafers based on defect pre-detection is disclosed in this embodiment. This embodiment focuses on the doping concentration testing of 4H-SiC epitaxial wafers. The testing process is optimized using the method described in this application, and the specific steps are as follows:

[0029] Step 1: Select 10 4H-SiC (150 mm in diameter, 10 μm in epitaxial layer thickness) epitaxial wafers. Each wafer is then subjected to acetone ultrasonic cleaning, isopropanol ultrasonic cleaning, and nitrogen drying in sequence. After cleaning, the wafers are observed with an optical microscope to confirm that there is no visible contamination on the surface.

[0030] Step 2: Use an optical defect detector (model: KLA Candela 8520) to perform a full-area scan on each sample with a scanning resolution of 1 μm, and mark areas such as triangular defects and scratches with a size greater than 5 μm; preset 17 initial test points, of which 1 point in each of the 3 samples is located at a distance of 25 μm from the triangular defect, which is identified as an anomaly and excluded;

[0031] Step 3: Redefine replacement points for the abnormal points of the above 3 samples. Select replacement points in the defect-free area around them. The distance between the replacement points and the original abnormal points is 100 μm, and the distance between the replacement points and other initial points is 10 mm. After confirming that there are no defects in the area where the replacement points are located by a second scan with a defect detector, synchronize the coordinates to the mercury probe CV testing system (model: Semilab MCV-530L).

[0032] Step 4: At an ambient temperature of 22 ℃ and a relative humidity of 50%RH, fix the sample on the stage. The mercury probe will contact the sample sequentially according to the coordinates of the points as required by the program. Control the same contact pressure and contact time, and obtain the CV curve of each point. Record the corresponding doping concentration and coordinate information.

[0033] Step 5: Test data for one surrogate site out of 10 samples (doping concentration 1.2 × 10¹). 6 The deviation of cm⁻³ from the average value of the standard samples in the same batch (1.0×10¹) 6 The percentage of 10% (cm⁻³) was considered abnormal and required secondary investigation. Steps two through four were repeated, and a rescan of the area revealed an unmarked 0.5 μm pit. A new replacement point was defined 150 μm away from the original replacement point, and the test result was 1.02 × 10¹. 6 cm⁻³, meets the requirements;

[0034] Implementation results: The average testing time for 10 samples was 15 min / sample, and the effective data rate was 100% without repeated testing. Compared with the traditional method of defect-free pre-inspection, the traditional method has an average testing time of 20 min / sample and an effective data rate of 85%. This embodiment improves efficiency by 33%, reduces the number of probe contacts by 30%, reduces the risk of mercury leakage, and the capillary lifespan is expected to be extended to 3 months.

[0035] Example 2:

[0036] A mercury probe CV testing method for silicon carbide epitaxial wafers based on defect pre-detection is disclosed in this embodiment. This embodiment is for testing the doping concentration of 4H-SiC epitaxial wafers with high defect density. For multi-site anomalies, the testing process is optimized using the method of this application. The specific steps are as follows:

[0037] Step 1: Select one high defect density 4H-SiC (diameter 150 mm, epitaxial layer thickness 10 μm, defect density ≥ 1 defect / cm²) epitaxial wafer. Each wafer is then subjected to acetone ultrasonic cleaning, isopropanol ultrasonic cleaning, and nitrogen drying in sequence. After cleaning, the wafer is observed with an optical microscope to confirm that there is no visible contamination on the surface, but the defect distribution is uniform, which meets the test conditions of "multi-point anomaly".

[0038] Step 2: Use an optical defect detector (model: KLA Candela 8520) to perform a full-area scan on each sample with a scanning resolution of 1μm, and mark areas such as triangular defects and scratches with a size greater than 5μm; preset 9 initial test points, of which 5 points of the initial points of the sample are located at a distance of 25μm from the triangular defects, which are identified as abnormal points and excluded;

[0039] Step 3: Redefine replacement points for the abnormal points of the above samples. Select replacement points within the defect-free area surrounding the abnormal points. The distance between the replacement points and the original abnormal points should be 100 μm, and the distance between the replacement points and other initial points should be 10 mm. A second scan using a defect detector confirms that two points in the area containing the replacement points are located 25 μm from the triangular defect. These are then identified as abnormal points and eliminated. Redefine replacement points again. Select replacement points within the defect-free area surrounding the abnormal points. The distance between the replacement points and the original abnormal points should be 150 μm, and the distance between the replacement points and other initial points should be 10 mm. After a third scan using the defect detector confirms the absence of defects, synchronize the coordinates to the mercury probe CV testing system (model: Semilab MCV-530L).

[0040] Step 4: At an ambient temperature of 22 ℃ and a relative humidity of 50%RH, fix the sample on the stage. The mercury probe will contact the sample sequentially according to the coordinates of the points as required by the program. Control the same contact pressure and contact time, and obtain the CV curve of each point. Record the corresponding doping concentration and coordinate information.

[0041] Implementation results: After optimization, the average doping concentration of the sample was 1.5 × 10¹. 6 The sample density was 1 cm⁻³, with a relative standard deviation of 1.7%, and no obvious mercury residue was observed on the surface. Under unoptimized conditions, the average doping concentration was 1.8 × 10¹⁰. 6 The sample size was cm⁻³, with a relative standard deviation of 5.2%, and significant mercury residue was observed on the surface. CV curve analysis showed that the optimized data had a validity rate of 100%, while the unoptimized data had a validity rate of 77.8%.

[0042] The above provides a mercury probe CV testing method for silicon carbide epitaxial wafers for defect pre-detection. This embodiment provides a mercury probe CV testing device for silicon carbide epitaxial wafers that can implement the above method, such as... Figure 2 The specific structure is as follows:

[0043] Optical Defect Detector: Through high-resolution optical imaging technology, it performs full-area scanning of the semiconductor wafer surface, accurately identifies various surface morphology defects, and transmits statistical information on defect locations to the control system in real time.

[0044] Wafer transfer robot: Based on defect detection results, it automatically picks up wafers and transfers them to the designated test area of ​​the mercury probe CV test module to achieve precise wafer transfer.

[0045] Mercury probe CV testing module: Equipped with intelligent defect reading and coordinate avoidance algorithms, including information input and analysis module, coordinate calibration module, defect area determination module, defect-free area planning module, and point verification and error correction module. After receiving the point information from the optical defect detector, it automatically identifies the defect coordinates at the original point, eliminates the defect area before contact, and redefines the test point, so that the mercury droplet only contacts the defect-free flat area of ​​the wafer. At the same time, it completes CV electrical testing through high-precision probe control technology and outputs relevant results.

[0046] It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of the invention. This method can adjust the defect detection threshold, detection point distribution, and test parameters according to the type of the sample to be tested. All equivalent variations and modifications made in accordance with the claims of this invention are within the scope of this invention.

Claims

1. A mercury probe CV testing method for silicon carbide epitaxial wafers based on defect pre-detection, characterized in that, The steps are as follows: Step 1: Prepare the test sample and clean the surface of the test sample to remove surface impurities and contaminants; Step 2: Use an optical defect detector to perform a full-area scan of the surface of the sample to be tested, identify the surface morphology defect areas, and mark the initially preset test points in the defect areas, excluding abnormal points located in the defect areas or ≤50 μm away from the defect areas; Step 3: For the excluded abnormal points, redefine alternative test points in the surrounding defect-free area. The distance between the alternative points and the original abnormal points should be 50-200 μm, and the spacing between the alternative points should not be less than twice the diameter of the test probe to avoid mutual interference of test signals. Step 4: Place the sample to be tested on the stage of the mercury probe CV test system, and control the mercury probe to contact each point in sequence according to the test points redefined in Step 3 to complete the electrical parameter test and record the test data. Step 5: If the test data of a certain alternative point in Step 4 still deviates from the normal range, and deviates from the average effective test value of the same batch by ≥10%, repeat Step 2 to Step 4 to eliminate the abnormality again and redefine the point until effective test data is obtained, and then continue to complete the mercury probe CV test of all samples to be tested.

2. The method for mercury probe CV testing of silicon carbide epitaxial wafers based on defect pre-detection according to claim 1, characterized in that: Surface morphology defects include fallen particles, carrot-like structures, step agglomerates, microtubules, triangular defects, scratches, and pits.

3. The method for mercury probe CV testing of silicon carbide epitaxial wafers based on defect pre-detection according to claim 1, characterized in that: The optical defect detector in step two includes optical defect detection equipment and a deep learning-based automatic defect detection algorithm. The scanning resolution is not less than 1μm. The number and corresponding coordinates of the identified feature defects must be marked, and the defect definition grid must include the maximum coverage area of ​​the defects.

4. The method for mercury probe CV testing of silicon carbide epitaxial wafers based on defect pre-detection according to claim 1, characterized in that: In step two, the spacing between adjacent initial points is set to 10 mm to avoid the edge area of ​​the test sample and deduct the edge range.

5. The method for mercury probe CV testing of silicon carbide epitaxial wafers based on defect pre-detection according to claim 1, characterized in that: The redefinition of the alternative point in step three needs to be confirmed in real time by the defect detector to ensure that there are no identifiable defects in the area where the current alternative point is located, and that the coordinates of the alternative point are automatically recorded by the software of the test system and synchronized to the mercury probe coordinate control module.

6. The method for mercury probe CV testing of silicon carbide epitaxial wafers based on defect pre-detection according to claim 1, characterized in that: In step four, the contact pressure and contact time of the mercury probe are kept constant, and the ambient temperature is maintained at 22±2℃ and the relative humidity is less than 50%RH during the test to avoid environmental factors affecting the test stability.

7. The method for mercury probe CV testing of silicon carbide epitaxial wafers based on defect pre-detection according to claim 1, characterized in that: The method can be integrated into existing mercury probe CV testing systems. By adding a defect detection module, a point optimization algorithm, and software integration, automated operation can be achieved without large-scale modifications to the original testing hardware.

8. A mercury probe CV testing apparatus for silicon carbide epitaxial wafers according to any one of claims 1-7, characterized in that, include: An optical defect detector is used to perform full-area scanning of the semiconductor wafer surface through high-resolution optical imaging, accurately identify various surface morphology defects, and transmit statistical information on defect locations to the control system in real time. The wafer transfer robot is used to automatically pick up wafers based on defect detection results and transfer them to the designated test area of ​​the mercury probe CV test module; Mercury probe CV testing module: Equipped with intelligent defect reading and coordinate avoidance algorithms, including information input and analysis module, coordinate calibration module, defect area determination module, defect-free area planning module, and point verification and error correction module. After receiving the point information from the optical defect detector, it automatically identifies the defect coordinates at the original point, eliminates the defect area from "before contact" and redefines the test point, so that the mercury droplet only contacts the defect-free flat area of ​​the wafer. At the same time, it completes the CV electrical test through probe control and outputs relevant results.

Images