Online data processing method and system

Abstract

The invention belongs to the technical field of semiconductor device test data processing. The invention provides an online data processing method and system, and the method comprises the steps: receiving an access request of a current client, and obtaining original test data based on the access request; the network module performs electrical parameter extraction on the original test data, and determines an image to be generated and sample information; based on the extracted electrical parameters, the determined to-be-generated image and the sample information, performing formatting processing by utilizing a customization conversion protocol, and based on the current client, determining a customization template matched with the current client so as to generate corresponding test reports in batches according to the determined customization template; and according to the input parameters of the current client, visually displaying the extracted electrical parameters and the generated test report. According to the method, the extracted electrical parameters and the generated test report are visually displayed, and the data analysis efficiency is improved by 5-20 times.

Description

Technical Field

[0001] The present invention relates to the technical field of semiconductor device test data processing, and in particular to an online data processing method and system. Background Art

[0002] In R&D and production, it is necessary to process and analyze the raw data obtained from various testers, extract key parameters, draw various images, and then manually summarize the obtained parameters and images into a report to determine what researchers really care about. With the development of large-scale automation, data has also expanded, and a lot of time and repetitive work need to be spent on data analysis and data collation and aggregation, which greatly reduces the efficiency of R&D and production. In addition, the results obtained are usually not uniform and standardized, making it difficult to make more advanced data comparisons.

[0003] There are many technical problems in the prior art. With the expansion of R&D and production scale, unified data processing is more conducive to the comparison of conditions and data, but it is difficult for a single-machine software script to be updated synchronously by everyone, which is not conducive to the promotion of standardization, resulting in confusion in software versions and high barriers to use. With the increase in integration, fields such as chip testing often face a large amount of data, which requires a lot of repetitive work and consumes a lot of manpower. In the process of extracting electrical parameters and drawing images, there are large subjective factors, which will affect the final extracted results and are not conducive to further analysis. In addition, it is difficult for test reports written manually to standardize and unify the format, which is not conducive to the horizontal comparison of multiple sets of data.

[0004] Therefore, it is necessary to provide a new method for automatically generating a test report of a semiconductor device to solve the above problems. Summary of the invention

[0005] The present invention aims to provide an online data processing method and system to solve the problems in the prior art that the software scripts of a single machine cannot be updated synchronously by all persons, which is not conducive to standardized promotion, leading to confusion in software versions and a high threshold for use. The amount of test data for semiconductor devices is huge, requiring a large amount of repetitive work and consuming a lot of manpower. There are large subjective factors in the process of extracting electrical parameters and drawing images, which leads to low accuracy of the extracted results. The format of the manually written test report is non-standard and non-uniform, and is even not conducive to horizontal comparison of multiple groups of data. The technical problems to be solved by the present invention are achieved through the following technical solutions.

[0006] The first aspect of the present invention proposes an online data processing method, which is applied to a server, and includes: receiving an access request from a current client, and obtaining original test data based on the access request; a network module extracts electrical parameters from the original test data, and determines an image and sample information to be generated; the electrical parameters include electrical parameters related to an IV curve, a transfer characteristic curve, and an inverter voltage transfer curve, and specifically include: extracting linear regions corresponding to different electrical parameters according to the characteristics of a target curve and the characteristics of a linear region to be extracted; determining a reference value or reference range of each electrical parameter according to the extracted linear region; formatting the extracted electrical parameters, the determined image and sample information to be generated, using a customized conversion protocol, and determining a customized template matching the current client based on the current client, so as to batch generate corresponding test reports according to the determined customized template; and visually displaying the extracted electrical parameters and the generated test report according to the input parameters of the current client.

[0007] The second aspect of the present invention provides an online data processing system, which executes the online data processing method described in the first aspect of the present invention, and the online data processing system includes: a receiving and processing module, which is used to receive an access request from a current client and obtain original test data based on the access request; a network module, which is used to extract electrical parameters from the original test data and determine the image and sample information to be generated; the electrical parameters include electrical parameters related to the IV curve, the transfer characteristic curve and the inverter voltage transfer curve, specifically including: extracting linear regions corresponding to different electrical parameters according to the characteristics of the target curve and the characteristics of the linear region to be extracted; determining the reference value or reference range of each electrical parameter according to the extracted linear region; a data analysis module, which performs formatting processing using a customized conversion protocol based on the extracted electrical parameters, the determined image and sample information to be generated, and determines a customized template matching the current client based on the current client, so as to batch generate corresponding test reports according to the determined customized template; a report generation module, which visually displays the extracted electrical parameters and the generated test report according to the input parameters of the current client.

[0008] The third aspect of the present invention provides an electronic device, comprising: one or more processors; a storage device for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the online data processing method described in the first aspect of the present invention.

[0009] A fourth aspect of the present invention provides a computer-readable medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the online data processing method described in the first aspect of the present invention.

[0010] The embodiments of the present invention include the following advantages: Compared with the prior art, the present invention receives an access request from the current client and obtains original test data based on the access request; the network module extracts electrical parameters from the original test data, determines the image and sample information to be generated, and can accurately extract electrical parameters; based on the extracted electrical parameters, the determined image and sample information to be generated, formatting is performed using a customized conversion protocol, and based on the current client, a customized template matching the current client is determined to batch generate corresponding test reports according to the determined customized template; according to the input parameters of the current client, the extracted electrical parameters and the generated test reports are visually displayed, which improves data analysis efficiency by 5 to 20 times without the need to install an online data processing system; and a visual graphical interface with simple operation is effectively realized.

[0011] In addition, by extracting electrical parameters from the original test data, the image and sample information to be generated are determined. When extracting electrical parameters, the linear regions corresponding to different electrical parameters are extracted according to the characteristics of the target curve and the characteristics of the linear region to be extracted. The electrical parameters can be accurately extracted and the reference value or reference range of each electrical parameter can be accurately determined. Based on the extracted electrical parameters, the determined image and sample information to be generated, a customized conversion protocol is used for formatting processing, and corresponding test reports are generated in batches according to customized templates. While standardizing the format of the original data, it is possible to efficiently realize automated and batch data analysis, and to obtain test reports intelligently without being affected by human intervention.

[0012] In addition, while ensuring the precision and accuracy of the extracted electrical parameters, the analysis results are more objective and comparable. It can extract electrical parameters and draw graphs based on the original data, thereby automatically generating a test report with a standardized format, visualizing the various contents of the test report, and effectively displaying the horizontal comparison results. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 is a flowchart of the steps of the online data processing method of the present invention; Figure 2 is a schematic diagram of an application example of the online data processing method of the present invention; Figure 3 It is a schematic diagram of an example of a smoothed curve obtained by smoothing original test data of current or voltage using a five-point smoothing moving algorithm in the online data processing method of the present invention; Figure 4 It is a schematic diagram of determining the intersection point of the reference line and the transfer characteristic curve in the online data processing method of the present invention; Figure 5 is a diagram showing an example of a process of extracting a subthreshold swing linear region; Figure 6 The following table shows the current limit of 0A, 10A and 10A. -9 A. 10 -8 A is an example of a curve comparison diagram of extracting the linear region of the subthreshold swing using the original curve and the curve after smoothing; Figure 7 It is a comparative schematic diagram of extracting the off-state current using the original curve and the optimized curve; Figure 8 It is a structural block diagram of the online data processing system of the present invention; Fig. 9 is a schematic structural diagram of an electronic device embodiment according to the present invention; Fig.10 It is a schematic diagram of the structure of a computer-readable medium embodiment according to the present invention. DETAILED DESCRIPTION

[0014] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0015] In view of the above problems, the present invention proposes an online data processing method, which receives an access request from a current client and obtains original test data based on the access request; a network module extracts electrical parameters from the original test data, determines the image and sample information to be generated, and can accurately extract electrical parameters; based on the extracted electrical parameters, the determined image and sample information to be generated, formatting is performed using a customized conversion protocol, and based on the current client, a customized template matching the current client is determined to batch generate corresponding test reports according to the determined customized template; according to the input parameters of the current client, the extracted electrical parameters and the generated test reports are visually displayed, which improves the data analysis efficiency by 5 to 20 times (specifically, before using the platform, other departments need 1 hour to 3 hours for manual analysis, while the online data processing system only needs about 5 minutes to 10 minutes to obtain complete analysis results, which reduces the processing time by about 5 to 20 times), without the need to install an online data processing system; and a visual graphical interface with simple operation is effectively realized.

[0016] Example 1 Refer to the following Figure 1 , Figure 2 , Figure 3 , Figure 4 , Figure 5 , Figure 6 and Figure 7 , the contents of the present invention will be described in detail.

[0017] Figure 1It is a flow chart of the steps of the online data processing method of the present invention.

[0018] like Figure 1 As shown, in step S101, an access request from the current client is received, and original test data is obtained based on the access request.

[0019] Specifically, the original test data refers to data files measured by various instruments. The original test data is obtained by being compatible with multiple types of tests. The compatible tests include IV curves, transfer characteristic curves, and inverter voltage transfer curves.

[0020] It should be noted that in other embodiments, the compatible tests may also be capacitance tests, output curve tests, hysteresis analysis, drain-induced barrier reduction, and repeated tests for analyzing device stability. It may also include hot carrier injection (HCI) tests and bias temperature instability tests (BTI). Normalized images may be further generated based on the data of these tests, or relevant parameters may be extracted. The above is only described as an optional example and cannot be understood as a limitation of the present invention.

[0021] Figure 2 FIG. 1 is a schematic diagram of an application example of the online data processing method of the present invention. Figure 2 As shown, in an application example, it specifically includes an online data processing system located at a server and a client that can interact with the server. Specifically, a network module of the online data processing system can interact with a user of the client. The network module can communicate with a storage module, a data analysis module, and a report generation module. The data analysis module is integrated with an algorithm for extracting electrical parameters.

[0022] Specifically, the test report includes test reports related to the performance of field effect transistors, inverters, resistors, and diodes. The electrical parameters include saturation current, off-state current, on / off ratio, transconductance, threshold voltage, drain-induced barrier lowering, and subthreshold swing.

[0023] In a specific embodiment, the network module can be deployed on a large server, a personal computer, a microcomputer and a mobile device, wherein the microcomputer is a small computer represented by Raspberry Pi, and the mobile device mainly refers to a mobile phone with an Android system. The client is, for example, a PC (the same computer can be used as a network module and a client at the same time), a mobile terminal and all other devices that support a browser. The client accesses the server through a browser to obtain an interactive interface, and uploads, for example, a data file to be processed on the interactive interface to change the relevant entry items of the extraction parameters. Then, after the server processes, for example, voltage and current data according to user requirements, the extracted electrical parameters, data processing calculation results, and related images are returned to the interactive interface, where the interactive interface is, for example, a web page on a browser.

[0024] For example, client a (i.e., the current client) sends an access request to an online data processing system, and obtains original test data based on the access request. When the online data processing system receives the access request, the server of the online data processing system parses the original test data in the access request, such as test items related to each test device, and returns the parsed original test data to client a.

[0025] The network module is the hub connecting each module. It accepts data from the client through the http protocol, stores the data in the storage module of the server, extracts, analyzes, plots, generates reports through the data analysis module, and sends the obtained data and reports to the client. Among them, the above-mentioned http protocol is called using the bottle module of python, and the storage module uses the os module of python to save and read files. The data analysis module is a series of algorithms independently developed based on pandas, scipy, and numpy. The matplotlib module used for image generation. The report generation module is based on PYPDF2 to render the data into a report in PDF format.

[0026] For example, the interactive data between the client and the server above uses echarts to render data images for generating line charts and histograms. Bootstrap implements responsive layout; vue is used as the main front-end control framework. The request to send data uses, for example, the POST request in the http protocol.

[0027] It should be noted that for users, there is no need to download and install any software, applications, or programs locally. As long as you access the server through a browser, you can use the above data analysis algorithm. The speed of data analysis depends on the speed of the server. A high-performance server can serve users in the entire LAN and achieve high-speed data analysis without relying on user devices. On the basis of a fixed data analysis module, the architecture can develop customized interaction methods suitable for different purposes to meet the needs of different users. In addition, for users who aim at data analysis, more modifiable parameters will be provided, such as file filters, excel worksheet filters, data xy axis definitions, etc. In terms of analysis results, the specific performance parameters of each device, as well as whether it fails and the reasons for failure will also be displayed. In addition, for users whose main demand is to generate electrical test reports, more standardized content will be specified, such as file naming specifications, excel worksheet names, etc. By uploading the data obtained by the standard test process, a complete test report can be obtained directly. For users who are more concerned about different batches, since the storage module stores files and reports according to certain specifications, it is also possible to extract reports generated at different times, and compare key data in the reports, such as on-state current, switching ratio, threshold voltage, subthreshold swing, etc., for different batches on different dates. The above are several examples of customized data analysis interactions. Users can obtain the most suitable customized analysis system according to their own needs and device characteristics. The above is only explained as an optional example and cannot be understood as a limitation of the present invention.

[0028] Next, in step S102, the network module extracts electrical parameters from the original test data to determine the image and sample information to be generated; the electrical parameters include electrical parameters related to the IV curve, the transfer characteristic curve and the inverter voltage transfer curve, specifically including: extracting linear regions corresponding to different electrical parameters according to the characteristics of the target curve and the characteristics of the linear region to be extracted; and determining the reference value or reference range of each electrical parameter according to the extracted linear region.

[0029] Electrical parameters are extracted from the original test data to determine the image and sample information to be generated; the electrical parameters include electrical parameters related to the IV curve, the transfer characteristic curve and the inverter voltage transfer curve.

[0030] Specifically, the original test data refers to data files measured by various instruments.

[0031] In a specific embodiment, a test module for batch automated data analysis is developed, which can be used to parse data files measured by various instruments and extract corresponding electrical parameters. Specifically, it supports the extraction of relevant electrical parameters of IV curves, transfer characteristic curves and inverter voltage transfer curves. The test module includes an extraction algorithm for extracting relevant electrical parameters.

[0032] Algorithms for extracting saturation current, off-state current, on / off ratio, threshold voltage, subthreshold swing, maximum transconductance, and hysteresis from a single transfer characteristic curve; algorithms for calculating drain-induced barrier reduction from multiple transfer characteristic curves; judging whether a device has failed based on the transfer characteristic curve, and the criteria for failure judgment include channel short circuit, channel open circuit, gate leakage, logic error, and subthreshold anomaly; algorithms for extracting threshold, gain, hysteresis, and noise margin from a single inverter voltage transfer curve; Multiple curves are extracted respectively, and the mean and standard deviation of the extracted multiple electrical parameters are calculated and analyzed.

[0033] In a specific embodiment, data analysis is performed on the original test data (multiple discrete data points, for example, x and y represented by an array with a length of ch) to form an original curve, and the intersection of the target curve (i.e., the target curve corresponding to the original curve) and the reference line (e.g., a straight line) is calculated.

[0034] Preferably, a five-point smoothing algorithm is used to smooth the original test data of the current or voltage to obtain a smoothed curve. Figure 3 Specifically, the value of each point after smoothing is equal to the average value of the point itself and the two points before and after it, a total of five points, where the first point and the last point are taken from themselves, and the second point and the fourth point are taken from the average value of the three points before and after it. Through the above-mentioned smoothing process, the influence of low current caused by accuracy fluctuation can be greatly reduced.

[0035] Further, the original test data (specifically the original curve) is subjected to electrical parameter extraction to determine the image and sample information to be generated, wherein the electrical parameters include electrical parameters related to the IV curve, the transfer characteristic curve and the inverter voltage transfer curve, specifically including the threshold voltage and subthreshold swing of the field effect transistor, the gain of the inverter, the conversion voltage, etc. Specifically, the original test data (including the table data, etc.) is input into the test module, and the corresponding image, i.e. the image to be generated, is automatically output, for example, the transfer characteristic curve of the field effect transistor, the inverter voltage transfer curve, the curve graph of the resistor element, the curve graph or histogram of the diode.

[0036] Furthermore, the sample information includes text information such as process name, device batch, test time, tester, test instrument, etc.

[0037] It should be noted that the above is only described as an optional example and should not be understood as a limitation to the present invention.

[0038] In an optional implementation, when extracting electrical parameters, linear regions corresponding to different electrical parameters are extracted according to the characteristics of the target curve and the characteristics of the linear region to be extracted, and further according to the extracted linear region.

[0039] Specifically, the target curve includes a transfer characteristic curve of a field effect transistor, a voltage transfer curve of an inverter, a curve graph of a resistor element, and a curve graph of a diode.

[0040] For carbon-based devices, when the target curve is a transfer characteristic curve of a field effect transistor, extracting linear regions corresponding to different electrical parameters specifically includes the following steps.

[0041] Step S201: derive the curve formed by the original test data and take the absolute value to find the reference point with the largest absolute value.

[0042] Specifically, before the curve formed by the original test data (i.e., the original curve) is derived, a smoothing process is performed first. For a schematic diagram of the principle of determining the reference point, see Figure 4 .

[0043] Furthermore, the original smoothed curve is differentiated and its absolute value is taken.

[0044] It should be noted that in the specifications of silicon-based devices, a straight line is directly drawn between the two points at the beginning and end of the linear region to calculate the relevant slope and intercept. However, the stability of carbon-based devices is not high enough at present, so linear fitting is required to accurately find the linear region. In addition, there are many fitting functions in Python, such as the linear fitting performed by the curve_fit function in the optimize library in the scipy module. This algorithm is similar to the least squares method in mathematics and can return the slope, intercept and correlation coefficient of the fitting. The above is only explained as an optional example and cannot be understood as a limitation of the present invention.

[0045] Step S202: repeatedly collect data points on both sides of the reference point until the number of data points is less than or equal to the specified number and a data point less than the specified value is encountered, and use the collected data points as a linear region to further perform linear fitting.

[0046] Specifically, data points on both sides of the reference point with the largest absolute value are collected until the number of data points is less than or equal to 4 (for example, the specified number is 4) and a data point less than 0.85 times the maximum absolute value is encountered, and the collected data points are used as the linear region for further linear fitting.

[0047] In this embodiment, the electrical parameters include saturation current, off-state current, on / off ratio, transconductance, threshold voltage, drain-induced barrier lowering, and subthreshold swing.

[0048] The saturation current is also called the on-state current. The maximum value in the transfer characteristic curve is selected as the on-state current, but the current at this time may not be the current of the device in the saturation region. Therefore, when extracting, it will be determined whether the position of the current is included in the linear region of the threshold voltage. If the linear region of the threshold voltage touches the beginning or end of the transfer characteristic curve, it means that the test range of the transfer characteristic curve does not include the saturation region, and a warning message will be provided to inform the user that the gate voltage scanning range needs to be modified.

[0049] It should be noted that when extracting electrical parameters, it is necessary to fit the linear region of a curve, and the search for the linear region is highly subjective. This subjectivity will cause the parameters extracted by different people or even by the same operator at different times to vary. Therefore, a customized method for finding the linear region that can be quantified and accurately reproduced is needed.

[0050] The linear regions corresponding to different electrical parameters are extracted according to the characteristics of the target curve and the characteristics of the linear region to be extracted.

[0051] For example, in linear coordinates, the slope of the transfer characteristic curve increases first and then decreases, and the linear region used to extract the threshold voltage appears near the point with the maximum slope. In logarithmic coordinates, the slope of the transfer characteristic curve also increases first and then decreases, but the position is different from that of the linear coordinate. Specifically, the linear region of the subthreshold swing is extracted at the turning point of the transfer characteristic curve or the point with the maximum curvature change.

[0052] For the linear region of threshold voltage extraction and subthreshold swing extraction (corresponding to Figure 5 In the "linear region" (in the "linear region"), the curves are from slow to steep and then slow again, specifically, the slope first increases and then decreases, so a part of both sides of the maximum slope can be taken as the linear region. In fact, it is also necessary to select an extraction coefficient (between 0 and 1) to determine the width of the linear region. In this embodiment, the extraction coefficient is in the range of 0 to 1, and can be selected as 0.85, that is, collecting from both sides of the maximum slope, stopping when the slope is less than 0.85 times the maximum slope, and all the collected points are taken as the linear region, and then linear fitting is performed to obtain the required electrical parameters. The specific process is as follows Figure 5 shown.

[0053] After smoothing (specifically derivative processing), some curves may still be mistakenly extracted into the off-state fluctuation region when extracting the linear region. In this case, the excision algorithm needs to be used to further search for the correct linear region. When the data length of the linear region is less than 4 points, the excision algorithm will be triggered. The specific excision algorithm is to set a number of limiting currents. Data below this current will be cut off and will not participate in the linear region extraction operation. For example, the limiting current is 0A, 10 -10 A. 10 -9 A. 10 -8 A. For details, see Figure 5 . Figure 5 The following table shows the current limit of 0A, 10A and 10A. -9 A. 10 -8 A curve comparison example of extracting the subthreshold swing linear region using the original curve and the smoothed curve. Specifically, different limiting currents are used in turn to extract the linear region (the curve selected within a pair of parallel dotted lines extending along the Y axis or numerical direction is the linear region), and the result with the longest linear region is selected as the linear region for further analysis. When the off-state current fluctuates greatly, the existing extraction method will cause the linear region to be incorrectly extracted to the off-state position (corresponding to Figure 6 The curve comparison diagram of the limited current of 0A, that is, Figure 6 The excision method, that is, setting a specific limiting current, not analyzing the data less than the limiting current, and re-extracting the linear region, can find the appropriate linear region (corresponding to Figure 6 The current limit is 10 -9 A. 10 -8 A curve comparison diagram, that is Figure 6 ). Figure 6 It can be seen that if the limiting current is too large, the extracted linear region will be shorter, increasing the fitting error. Therefore, by comparing the linear regions extracted by different limiting currents, the optimal linear region can be determined, that is, the data with the longest linear region can be selected.

[0054] It should be noted that for the extraction of linear regions, influencing factors include extraction coefficients, excision algorithms, etc., which will affect the extraction results. The above-mentioned 0.85 and 4-point excision methods are used. However, in practice, regular updates and optimizations can be performed based on the evaluation results to make the extraction results more reasonable. In addition, it is impossible for an actual tester to collect countless points and can only use enough discrete data points to approximate a curve. Therefore, when using a computer to process some mathematical concepts, there will be different processing methods. In addition, for the limited current, in his implementation, other values ​​can also be selected. The above is only explained as an optional example and cannot be understood as a limitation of the present invention.

[0055] According to the extracted linear region, the reference value or reference range of each electrical parameter is determined.

[0056] In a specific embodiment, a reference value or a reference range of each electrical parameter is determined from a linear region extracted from a transfer characteristic curve of a field effect transistor.

[0057] Specifically, the saturation current, off-state current, switch ratio, transconductance, threshold voltage, drain-induced barrier lowering, and subthreshold swing are extracted from the linear region extracted from the transfer characteristic curve of the field effect transistor. Among them, the saturation current is the current of the transistor in the saturation state. The off-state current is considered to be the current when the device is turned off. The switch ratio is used to measure the switching ability of the device. The transconductance is used to measure the effect of the change in gate voltage on the change in current.

[0058] For example, the maximum value of the above transfer characteristic curve is directly taken as the saturation current. However, this method will encounter the problem that the working voltage does not cover the saturation region, so after extracting the saturation current, extract the linear region of the transfer characteristic curve to determine whether the extracted linear region will touch the beginning or end of the transfer characteristic curve. If it touches the beginning or end of the transfer characteristic curve, it is determined that the beginning or end of the transfer characteristic curve has not reached the saturation region, and a warning message will be given. For example, the minimum value of the transfer characteristic curve is taken as the off-state current, but this method will cause the current obtained to be too low when the off-state is very low due to the current fluctuation of the tester, resulting in failure to truly reflect the shutdown capability of the device. See Figure 7 , extracting the off-state current selects the minimum value of the current, but when the current is small, the test fluctuation will produce a signal that is 1 to 2 orders of magnitude lower than the actual off-state, and such an off-state current cannot truly reflect the shutdown capability of the device. In order to truly reflect the shutdown capability of the device, the present invention optimizes the method of obtaining the off-state current, first smoothes the transfer characteristic curve, and then uses the minimum value after smoothing as the off-state current, so that a more accurate off-state current can be obtained, and an off-state current that is more in line with the actual situation can be obtained, thereby accurately determining the shutdown capability of the device.

[0059] For example, the on / off ratio is calculated by taking the base 10 logarithm of the saturation current divided by the off-state current: OOR=log(I on / I off ) Where, OOR represents the on-off ratio; I on Indicates saturation current; I off Represents the off-state current.

[0060] Since the acquisition of OOR involves pure mathematical calculations, its accuracy is only affected by the extracted saturation current and off-state current.

[0061] For transconductance, the following expression is used to calculate the transconductance: G m = dI ds / dV gs Among them, G m Represents transconductance, which is used to measure the effect of gate voltage changes on current changes; V gs Represents gate voltage; I ds Indicates current.

[0062] Specifically, by finding the slope on the above transfer characteristic curve, the transconductance at each point can be obtained. The transconductance here refers to the maximum value of all transconductances.

[0063] In addition, the threshold voltage describes the transition point between the on and off states of a field effect transistor. In silicon-based devices, the physical meaning is the gate voltage (specifically Vgs) at which the hole-electron density at the channel interface is equal. Different extraction methods are used to extract the threshold voltage for different devices.

[0064] For example, the threshold voltage of carbon-based devices can be extracted by using the characteristics of threshold voltage. The linear reverse extension method is used, and the maximum transconductance method is used to find a linear region near the maximum transconductance. The intersection of the straight line fitted in this linear region and the horizontal axis is used as the threshold voltage. This extraction method is suitable for situations where the source-drain voltage is small (for example, V ds <0.5V). However, the printed electronics method first squares the current and then uses the linear reverse extension method to extract the threshold voltage. This extraction method is suitable for devices with large source-drain voltages (such as V ds >0.5V).

[0065] For example, for a batch of devices with relatively stable performance, the reference line method can be used to define a reference current (usually 10 -7 A / μm×channel width), take the transfer characteristic curve and the reference line (such as I ds =10 -8 The horizontal axis of the intersection of A) is taken as the threshold voltage. Due to the poor stability of carbon-based devices, the performance of devices varies greatly, and the off state of some devices cannot even reach 10 -8 A, therefore, a variety of extraction methods are used according to actual conditions.

[0066] In the above reference line method, an algorithm for finding the intersection of the transfer characteristic curve and the reference line is determined. The specific algorithm is to start from the first point and judge (y n -I ds )×(y n+1 -I ds ) is positive or negative, where y n is the ordinate of the current point, y n+1 is the ordinate of the next point, Ids The current value selected for the reference line. Find the first (y n -I ds )×(y n+1 -I ds ) is a negative value, the current point is taken as point A, and the next point of the current point is taken as point B. Figure 4 As shown, point A (x1, y1) or point B (x2, y2) can be directly used as the intersection point of the two. In order to further improve the accuracy, mathematical methods can be used to obtain the intersection point C (x0, y0) of the straight line passing through points AB and the reference line as the intersection point of the two (see Figure 4 The coordinates of point C are calculated as follows:

[0067]

[0068] This method can also be used to extract the intersection point between the hysteresis calculation reference line and the transfer characteristic curve.

[0069] Optionally, different methods are used to extract the off-state current of the above transfer characteristic curve, where the original curve is the result of the direct minimum value method, and the smoothed curve is the result of the minimum value after smoothing. From the results, it can be seen that for devices with relatively stable off-state, the off-state current will also be reduced by 2 times due to fluctuations; for devices with strong bipolarity, the above two methods are not much different; when the tester accuracy is low, the off-state will be reduced by 1~3 orders of magnitude. The data information such as the comparison of the above methods can be automatically generated as the content of the test report.

[0070] For example, according to the extracted linear region corresponding to the subthreshold swing, data points corresponding to the specified current and the specified voltage are determined to obtain a slope, and the inverse of the obtained slope is used as the subthreshold swing.

[0071] It should be noted that subthreshold swing (also known as S factor) is an important parameter for measuring the transistor turn-off performance when MOSFET operates in the subthreshold state.

[0072] For batches of devices with relatively stable performance, two points in the subthreshold region can be selected according to the specified current or voltage to calculate the subthreshold swing. Due to the poor stability of carbon-based devices, the threshold voltage extraction method is used for extraction. After taking the logarithm of the current, the point near the maximum slope is searched as the linear region, and a linear fit is performed, and the inverse of the slope is used as the subthreshold swing.

[0073] It should be noted that the subthreshold swing will increase with the increase of source-drain voltage, and this increase does not affect the extraction of the subthreshold swing by the program. However, during the R&D and production process, it is necessary to select the subthreshold swing under the same source-drain bias for comparison. For example, the transfer characteristic curve with a smaller Vds is selected for extraction to obtain the subthreshold swing.

[0074] Further, based on the electrical parameters extracted in the linear region, the following device failure causes are determined: gate leakage, channel short circuit, device judgment, logic field, and curve error.

[0075] In another optional embodiment, for a carbon-based device, when the target curve is a transfer characteristic curve of an inverter, the gain, hysteresis, and noise margin of the inverter are extracted, and noise analysis is performed.

[0076] It should be noted that the above is only described as an optional example and should not be understood as a limitation to the present invention.

[0077] Next, in step S103, based on the extracted electrical parameters, the determined image to be generated and the sample information, formatting is performed using a customized conversion protocol, and based on the current client, a customized template matching the current client is determined to batch generate corresponding test reports according to the determined customized template.

[0078] Specifically, based on the original test data, according to the defined X-axis, Y-axis, and the coordinate axis names corresponding to the X-axis and Y-axis, a curve graph of the transfer characteristic curve is generated, and I is extracted from the generated transfer characteristic curve. on ,I off , OOR, G m 、V th , SS, and then further generate corresponding graphics.

[0079] For example, a curve graph of the transfer characteristic curve is generated by directly taking GateV as the x-axis and DrainI or SourceI as the y-axis.

[0080] Optionally, when you draw a graph with gate voltage (Vgs) as the x-axis and drain current (Ids) or source current (Isrc) as the y-axis, you are actually creating an important graph for describing the behavior of a field effect transistor (FET), that is, the final graph (i.e., the generated image) from the original curve formed by the original test data to the target curve. This graph is very important for understanding the switching characteristics and amplification characteristics of the device.

[0081] The electrical parameters extracted in step S103, the image corresponding to the target curve, and the sample information (specifically including text information such as process name, device batch, test time, tester, and test instrument) are formatted using a customized conversion protocol.

[0082] Specifically, the customized conversion protocol is formed into source codes containing different STEXs, so that it is compatible with the parameter extraction and analysis protocols of the following test devices: field effect transistors, inverters, resistor elements, and diodes.

[0083] Optionally, the customized transformation protocol is embedded into the data analysis system as a plug-in.

[0084] The custom conversion protocol can extract various parameters from the original data (data folder) (various parameters are formatted and saved in output / report.stex), generate images from the original data (output folder), and finally parse them by the test module integrated with the custom conversion protocol (starTex folder) to automatically generate a test report (output / report.pdf). Specifically, the example.py of generating pdf from data to stex file, parsed source code.stex, parsed source code-compiled result.pdf, generated graphics (including histograms, line graphs, curve graphs, and corresponding code examples.py.

[0085] The test module specifically parses the file elements in the Doc list and renders them. When rendering, it reads each element in the Doc list in turn and renders them into a corresponding format. The Doc list contains serialized content and format. The file elements include multi-level titles, texts corresponding to each level of titles, tables, and images.

[0086] It should be noted that the Doc list is a custom conversion protocol (specifically, the most important part of the STEX source code). The test module parses the serialized content and format in the Doc list in turn and renders it into the corresponding page. The serialized content can be various file elements including multi-level titles, text, tables, images, paging, etc. The format includes font, font size, color, alignment, etc.

[0087] For example, the extracted language is Python, and the customized conversion protocol is in Python format, but its source code is in a highly compatible JSON format and can be embedded to collaborate with other languages.

[0088] There may be different specifications for different users, and customized conversion protocols can play a role in promoting a certain specification within a range.

[0089] In addition, files in formats such as stex and py can be opened using vscode.

[0090] For batch automatic generation, a loop is used to achieve one-click generation of multiple reports. Currently, the python language is used, but due to the high compatibility of the customized conversion protocol, it can be embedded in various systems to generate reports. You can also use php to call python to use the customized conversion protocol.

[0091] Furthermore, corresponding test reports are generated in batches according to customized templates, where the customized templates are, for example, user-customized templates corresponding to different device tests and different processes, and specifically include a cover, title, directory, text, table, and image.

[0092] For example, it can directly generate PDF (to achieve a PDF format that is easier for people to read than a computer-readable PDF), and define the styles of all elements in the PDF, including size, alignment, etc.

[0093] For example, choose an appropriate scale and ensure that the x-axis and y-axis are in appropriate proportions to clearly show data trends. Label key points, such as the location of important parameters such as threshold voltage and maximum current.

[0094] For example, using multiple curves, if you need to show the behavior under different conditions, you can plot multiple curves on the same graph, such as different temperatures or different Vds values. It also includes explanations of different regions and key features in the graph, such as saturation region, linear region, etc. It also includes parameter analysis text information.

[0095] For the images to be generated (corresponding to the target curve), they can also include transfer characteristics, characteristics, transconductance plots, subthreshold swing plots, and histograms. Among them, the transfer characteristics plot is used to show the relationship between the gate-source voltage and the drain current, helping to identify the threshold voltage and other important parameters. The characteristics plot refers to the relationship between the drain-source voltage and the drain current, especially under different gate-source voltages. The transconductance plot refers to a graph that plots the change of transconductance with the gate-source voltage. The subthreshold swing plot shows the performance in the subthreshold region. Histograms are used for statistical distribution, such as the threshold voltage distribution of multiple samples. In these graphs, the gate-source voltage (V gs ), and then change the drain-source voltage (V ds ), to observe the drain current (I ds). The curves corresponding to these figures are divided into three regions: cutoff region, saturation region and linear / ohmic region.

[0096] By automatically generating charts and text in the test report, various performance test analysis results of semiconductor devices can be better visualized.

[0097] The way to generate reports can be flexibly adjusted for different customers, different device types, and different test contents. In terms of data processing, various types of algorithms can be combined according to the data characteristics and the parameters to be extracted to obtain the required parameters. In terms of image generation, it can support multiple presentation methods such as line graphs, scatter graphs, histograms, heat maps, box plots, etc. to achieve intuitive and clear results. The final report type can also be typeset into various types as needed, including complete reports for a single device, comparative reports for multiple sets of data, etc., so that users can grasp more complete information.

[0098] The above algorithms can be various data processing methods such as curve cutting, curve smoothing, derivation, linear fitting, intersection point finding, intercept finding, basic statistics, etc. Users can combine these algorithms in various ways to extract target parameters. In addition, automatic selection items are also provided to automatically select algorithm combinations based on preset keywords.

[0099] Optionally, a large language model is used to describe data processing requirements to automatically generate algorithm combinations. Optionally, it is compatible with users writing their own data processing algorithms to achieve high customization. The most effective extraction method can be combined according to practical scenarios and user needs.

[0100] In a specific implementation, the threshold voltage may be extracted by extracting the linear region and then extending it in the reverse direction to obtain the intercept of the x-axis.

[0101] Optionally, a reference current i is set, and the abscissa of the intersection of the transfer characteristic curve and y=i is calculated as the threshold voltage.

[0102] For the subthreshold swing, find the slope of the logarithmic coordinate, and the logarithm of the slope is the subthreshold swing.

[0103]

[0104] Where SS is the subthreshold swing, V g is the gate voltage, I ds is the source-drain current.

[0105] In another embodiment, for the case of drain-induced barrier lowering (DIBL), different source-drain voltages (V ds) at the threshold voltage (V th ), and then calculate the slope of the threshold voltage changing with the source-drain voltage:

[0106] in, Indicates the slope of the threshold voltage as the source-drain voltage changes; in order to obtain DIBL, the source-drain voltage (V ds ) Two groups of test conditions, higher and lower, were used to test the transfer curve. Respectively represent the selected higher and lower source-drain voltages; It represents the threshold voltage extracted under the test condition of high source-drain voltage; Represents the threshold voltage extracted under test conditions where the source-drain voltage is low.

[0107] The linear region reverse extension method is used to extract the threshold voltage (hereinafter referred to as the linear reverse extension method). However, this method is more accurate when the source-drain voltage is low (less than 0.5V), and the extraction error is larger when the source-drain voltage is high (greater than 1.5V). For the transfer characteristic curve when the source-drain voltage is high, the square root of all currents is obtained ( ,in Represents the source-drain current in the transfer curve), then find the linear area, perform linear fitting, and sum the intercepts of the x-axis as the threshold voltage (because taking the square root is equivalent to taking the 0.5th power, it is also called the point-five method). Specifically select a set of continuous source-drain voltages (such as 0.1V, 0.5V, 1.0V, 1.5V, 2.0V) to test the transfer curve, extract the threshold voltages for linear fitting to get a more accurate DIBL, but due to the accuracy of the threshold voltage extraction mentioned above, it is necessary to simplify the DIBL and select only the data when the source-drain voltage is 0.1V and 2.0V. For data with a source-drain voltage of 2.0V, the point-five method is used to extract the threshold voltage, and for data with a source-drain voltage of 2.0V, the linear reverse extension method is used to obtain a more accurate DIBL value. According to different devices and actual conditions, other source-drain voltages can also be selected for DIBL extraction.

[0108] In another specific embodiment, the reference line method is used for extracting hysteresis, that is, the algebraic mean or geometric mean of the maximum and minimum values ​​of the data is used as the reference line, that is, or ,in, is the maximum value of the source-drain current in the transfer characteristic curve, is the minimum value of the source-drain current in the transfer characteristic curve. The x-coordinate difference of the intersection of the reference line with the outgoing line and the return line is taken as the hysteresis value, that is, ,in are the horizontal coordinates of the intersection of the reference line, the outgoing line and the return line. This method of extracting hysteresis can effectively avoid the problem of large differences in the curve loop at the reference line in the hysteresis analysis of the transfer characteristic curve. When the hysteresis phenomenon is obvious, the difference in threshold voltage can be used as the hysteresis value, that is, ,in The threshold voltages extracted using the removal and return lines are respectively. The threshold voltage can better reflect the essence of device performance. The hysteresis value extracted using this algorithm can more accurately reflect the hysteresis phenomenon of the device. Similarly, for the voltage transfer curve of the inverter, the threshold voltage is used to reflect the hysteresis phenomenon of the inverter, that is, ,in are the threshold voltages of the inverter curves extracted using the de-line and the re-line, respectively.

[0109] In another embodiment, if the minimum value is directly used as the off-state current according to the traditional method, the off-state current may be extracted incorrectly due to large off-state fluctuations. -11 A, while the off state of the device under test with good performance is difficult to reach 10 -10 A. In order to eliminate the influence of this situation, a five-point smoothing algorithm is used, that is, each point is replaced by the average value of the five points including the point and the two points before and after it. The first point and the fifth point are taken as the average value of themselves, and the second point and the fourth point are taken as the average value of themselves and the points before and after them. yy(d) represents the value of the dth point, and d is a positive integer, specifically 1, 2, ..., n.

[0110] For example, yy(1) = y(1) , yy(2) = (y(1) + y(2) + y(3)) / 3 , yy(3) = (y(1) + y(2) + y(3) + y(4) + y(5)) / 5 , yy(4) = (y(2) + y(3) + y(4) + y(5) + y(6)) / 5 .

[0111] By calculating the values ​​of the above points in sequence and taking the minimum value of the smoothed curve, a more reasonable off-state current can be obtained.

[0112] The above examples can fully illustrate that within the framework of the present invention, users can use existing, recommended algorithms, and custom algorithms to freely combine them, thereby extracting the most effective parameters and achieving scientific automated data analysis.

[0113] On the basis of extracted parameters, the yield of the device, performance fluctuations, and the overall situation of the process batch can be preliminarily judged and evaluated based on the preset parameters and the results of large language model learning. For example, the change in yield can be statistically analyzed to analyze the repeatability and time stability of the process.

[0114] The above-mentioned automatic report generation method includes automatically determining the data analysis method, image display method and report layout method to be used according to the file name and data features. Among them, the automatic judgment rule is specifically determined according to pre-set keywords (such as voltage parameters, hys, etc.). Optionally, it is inferred by a large language model. Users can make customized modifications based on automatic judgment to form a complete analysis plan. A complete analysis plan can be exported as a configuration file for repeated use or migration.

[0115] In one embodiment, the data analysis method is determined based on the pre-set keywords. Specifically, when analyzing the test data, GateV, VG, and Vgs will be identified as gate voltages, and DrainI, ID, and Ids will be identified as elemental drain currents. When selecting the data analysis method, files containing hys will be subjected to hysteresis analysis, and files containing DIBL will be subjected to drain-induced barrier reduction analysis; when selecting the generated image, data with position coordinates will be subjected to heat map analysis, multiple groups of data will be subjected to histogram analysis, and multiple batches of data will be subjected to box plot analysis.

[0116] In another embodiment, the prediction model constructed based on the large language model refers to collecting user information data, test equipment data used by the user, preferred image types, test items, test report formats, and saving habits, and marking the above data with preference items and report types, so that the prediction model can learn the relationship between the above data. When user information data is input, the corresponding test report template can be automatically generated according to the information items preferred by the user.

[0117] In another embodiment, a data analysis solution is generated more accurately based on the user's saving habits, the analysis content to be performed, and the pre-set keywords. The above configuration file can be a configuration text document in json or yaml format, which can be used for reading and writing of this module and can also be compatible with other analysis software for further analysis required by customers.

[0118] It should be noted that the above is only described as an optional example and should not be understood as a limitation to the present invention.

[0119] Next, in step S104, the extracted electrical parameters and the generated test report are visually displayed according to the input parameters of the current client.

[0120] In a specific implementation, the user configures a file filter, and uses the configured file filter to select a portion of the data files from the uploaded data files, and selects a portion of the data tables from the data files for analysis. For example, it is selected whether to perform failure analysis, and failure conditions are configured as needed, and corresponding parameters are input according to the selected test items, and parameters corresponding to the failure conditions are input according to the configured failure conditions. The server receives the input parameters, returns display data corresponding to the input parameters, and uses the display data to visually display the extracted electrical parameters and the generated test report on the interactive interface.

[0121] It should be noted that in this example, the user can select the x-axis display range of the line graph and the histogram; the user can also fill in the report-related information, such as product name, chip name, tester, test equipment, etc. These test items can be freely increased or decreased as needed. The above is only used as an optional example for explanation and cannot be understood as a limitation of the present invention.

[0122] In addition, the drawings are only schematic illustrations of the processes included in the method according to an exemplary embodiment of the present invention, and are not intended to be limiting. It is easy to understand that the processes shown in the drawings do not indicate or limit the time sequence of these processes. In addition, it is also easy to understand that these processes can be performed synchronously or asynchronously, for example, in multiple modules.

[0123] Compared with the prior art, the present invention extracts electrical parameters from original test data to determine the image and sample information to be generated. When extracting electrical parameters, the linear regions corresponding to different electrical parameters are extracted according to the characteristics of the target curve and the characteristics of the linear region to be extracted. The electrical parameters can be accurately extracted and the reference value or reference range of each electrical parameter can be accurately determined. Based on the extracted electrical parameters, the determined image and sample information to be generated, a customized conversion protocol is used for formatting, and corresponding test reports are generated in batches according to customized templates. While standardizing the format of the original data, efficient automated and batch data analysis can be achieved, and test reports can be obtained intelligently without being affected by human intervention.

[0124] In addition, while ensuring the precision and accuracy of the extracted electrical parameters, the analysis results are more objective and comparable. It can extract electrical parameters and draw graphs based on the original data, thereby automatically generating a test report with a standardized format, visualizing the various contents of the test report, and effectively displaying the horizontal comparison results.

[0125] Example 2 The following is a system embodiment of the present invention, which can be used to implement the method embodiment of the present invention. For details not disclosed in the system embodiment of the present invention, please refer to the method embodiment of the present invention.

[0126] Figure 8 It is a schematic structural diagram of an example of an online data processing system according to the present invention.

[0127] The second aspect of the present disclosure provides an online data processing system 800, which adopts the online data processing method described in the first aspect of the present disclosure.

[0128] like Figure 8 As shown, the online data processing system 800 includes a receiving and processing module 810 , a network module 820 , a data analysis module 830 and a report generation module 840 .

[0129] In a specific implementation, the receiving and processing module 810 is used to receive an access request from the current client, and obtain the original test data based on the access request. The network module 820 is used to extract electrical parameters from the original test data, and determine the image and sample information to be generated; the electrical parameters include electrical parameters related to the IV curve, the transfer characteristic curve, and the inverter voltage transfer curve, specifically including: extracting linear regions corresponding to different electrical parameters according to the characteristics of the target curve and the characteristics of the linear region to be extracted; according to the extracted linear region, determining the reference value or reference range of each electrical parameter. The data analysis module 830 uses the customized conversion protocol to format the extracted electrical parameters, the determined image and sample information to be generated, and determines the customized template matching the current client based on the current client, so as to batch generate the corresponding test report according to the determined customized template. The report generation module 840 visualizes the extracted electrical parameters and the generated test report according to the input parameters of the current client.

[0130] According to an optional implementation, the network module can communicate with the storage module, the data analysis module, and the report generation module, and the data analysis module is integrated with an algorithm for extracting electrical parameters; the test report includes a test report related to the performance of field effect transistors, inverters, resistors, and diodes; the electrical parameters include saturation current, off-state current, switching ratio, transconductance, threshold voltage, drain-induced barrier lowering, and subthreshold swing.

[0131] According to an optional implementation manner, the algorithm for extracting electrical parameters includes: an algorithm for extracting saturation current, off-state current, switching ratio, threshold voltage, subthreshold swing, maximum transconductance, and hysteresis from a single transfer characteristic curve; an algorithm for calculating drain-induced barrier reduction from multiple transfer characteristic curves; judging whether the device has failed based on the transfer characteristic curve, and the criteria for failure judgment include channel short circuit, channel open circuit, gate leakage, logic error, and subthreshold anomaly; an algorithm for extracting threshold, gain, hysteresis, and noise tolerance from a single inverter voltage transfer curve; extracting multiple curves separately, and calculating and analyzing the mean and standard deviation of the extracted multiple electrical parameters.

[0132] According to an optional implementation, the customized conversion protocol is formed into source code containing different STEXs, so that it is compatible with the parameter extraction and analysis protocols of the following test devices: field effect transistors, inverters, resistor elements, diodes; the customized conversion protocol is used as a plug-in and embedded in the data analysis module.

[0133] According to an optional implementation, the algorithm for extracting electrical parameters is called through the data analysis module to extract, analyze, plot and generate reports for the data, and the obtained data and reports are sent to the client to achieve visual display.

[0134] According to an optional implementation method, the prediction model constructed based on the large language model refers to collecting user information data, test equipment data used by the user, image types generated by preferences, test items, test report formats, and saving habits, and marking preference items and report types for the above data, so that the prediction model can learn the relationship between the above data. When user information data is input, the corresponding test report template can be automatically generated according to the information items preferred by the user.

[0135] According to an optional implementation manner, according to the device batches in the sample information, the information is parsed, extracted, and drawn in sequence to generate a plurality of different source code files for generating corresponding test reports; the sample information includes process name, device batch, test time, tester, and test instrument; the file elements in the Doc list are specifically parsed and rendered, and when rendering, each element in the Doc list is read in sequence and rendered into a corresponding format, wherein the file elements include multi-level titles, and texts, tables, and images corresponding to each level of titles.

[0136] According to an optional implementation manner, the step of extracting linear regions corresponding to different electrical parameters according to the characteristics of the target curve and the characteristics of the linear region to be extracted includes: For carbon-based devices, when the target curve is the transfer characteristic curve of a field effect transistor, extracting linear regions corresponding to different electrical parameters includes: taking the derivative of the curve formed by the original test data and taking the absolute value to find the reference point with the largest absolute value; repeatedly collecting data points on both sides of the reference point until the number of data points is less than or equal to the specified number and a data point less than the specified value is encountered, and using the collected data points as the linear region for further linear fitting; wherein the electrical parameters include saturation current, off-state current, switching ratio, transconductance, threshold voltage, drain-induced barrier lowering, and subthreshold swing.

[0137] It should be noted that the online data processing method executed by the online data processing system in this embodiment is substantially the same as the online data processing method in Embodiment 1, and therefore, the description of the same parts is omitted.

[0138] Compared with the prior art, the present invention extracts electrical parameters from original test data to determine the image and sample information to be generated. When extracting electrical parameters, the linear regions corresponding to different electrical parameters are extracted according to the characteristics of the target curve and the characteristics of the linear region to be extracted. The electrical parameters can be accurately extracted and the reference value or reference range of each electrical parameter can be accurately determined. Based on the extracted electrical parameters, the determined image and sample information to be generated, a customized conversion protocol is used for formatting, and corresponding test reports are generated in batches according to customized templates. While standardizing the format of the original data, efficient automated and batch data analysis can be achieved, and test reports can be obtained intelligently without being affected by human intervention.

[0139] In addition, while ensuring the precision and accuracy of the extracted electrical parameters, the analysis results are more objective and comparable. It can extract electrical parameters and draw graphs based on the original data, thereby automatically generating a test report with a standardized format, visualizing the various contents of the test report, and effectively displaying the horizontal comparison results.

[0140] Fig. 9 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

[0141] like Fig. 9 As shown, the electronic device is presented in the form of a general computing device. The processor may be one or more and work in coordination. The present invention does not exclude distributed processing, that is, the processor may be dispersed in different physical devices. The electronic device of the present invention is not limited to a single entity, but may also be the sum of multiple physical devices.

[0142] The memory stores a computer executable program, which is usually a machine-readable code. The computer-readable program can be executed by the processor to enable the electronic device to perform the method of the present invention, or at least part of the steps in the method.

[0143] The memory includes a volatile memory, such as a random access memory unit (RAM) and / or a cache memory unit, and may also be a non-volatile memory, such as a read-only memory unit (ROM).

[0144] Optionally, in this embodiment, the electronic device further includes an I / O interface, which is used for the electronic device to exchange data with an external device. The I / O interface can represent one or more of several types of bus structures, including a storage unit bus or a storage unit controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of a variety of bus structures.

[0145] It should be understood that Fig. 9The electronic device shown is only an example of the present invention, and the electronic device of the present invention may also include elements or components not shown in the above examples. For example, some electronic devices also include display units such as display screens, and some electronic devices also include human-computer interaction elements such as buttons, keyboards, etc. As long as the electronic device can execute the computer-readable program in the memory to implement the method of the present invention or at least part of the steps of the method, it can be considered as an electronic device covered by the present invention.

[0146] Through the above description of the implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by combining software with necessary hardware. Fig.10 As shown, the technical solution according to the embodiment of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes a number of commands to enable a computing device (which can be a personal computer, a server, or a network device, etc.) to execute the above method according to the embodiment of the present invention.

[0147] The software product may be any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, a system, device or component of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0148] The computer readable storage medium may include a data signal propagated in a baseband or as part of a carrier wave, wherein a readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The readable storage medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with a command execution system, device, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the above.

[0149] Program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0150] The computer-readable medium carries one or more programs (eg, computer-executable programs). When the one or more programs are executed by a device, the computer-readable medium implements the method of the present disclosure.

[0151] Those skilled in the art will appreciate that the above modules can be distributed in the device according to the description of the embodiment, or can be changed accordingly and only used in one or more devices different from the embodiment. The modules of the above embodiments can be combined into one module, or further divided into multiple sub-modules.

[0152] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the exemplary embodiments described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the embodiment of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several commands to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the embodiment of the present invention.

[0153] The exemplary embodiments of the present invention are specifically shown and described above. It should be understood that the present invention is not limited to the detailed structure, configuration or implementation method described herein; on the contrary, the present invention is intended to cover various modifications and equivalent configurations included in the spirit and scope of the appended claims.

Claims

1. An online data processing method, applied to a server, characterized in that: It includes Receive an access request from the current client, and obtain original test data based on the access request; The network module extracts electrical parameters from the original test data and determines the image and sample information to be generated; The electrical parameters include electrical parameters related to the IV curve, the transfer characteristic curve and the inverter voltage transfer curve, and specifically include: extracting linear regions corresponding to different electrical parameters according to the characteristics of the target curve and the characteristics of the linear region to be extracted; determining the reference value or reference range of each electrical parameter according to the extracted linear region; Based on the extracted electrical parameters, the determined images to be generated and the sample information, a customized conversion protocol is used for formatting, and based on the current client, a customized template matching the current client is determined to batch generate corresponding test reports according to the determined customized template; Based on the input parameters of the current client, the extracted electrical parameters and the generated test report are displayed visually.

2. The online data processing method according to claim 1, characterized in that: include: The network module can communicate with the storage module, the data analysis module, and the report generation module, and the data analysis module is integrated with an algorithm for extracting electrical parameters; Test reports include those related to the performance of field effect transistors, inverters, resistor elements, and diodes; The electrical parameters include saturation current, off-state current, on / off ratio, transconductance, threshold voltage, drain-induced barrier lowering, and subthreshold swing.

3. The online data processing method according to claim 2, characterized in that: The algorithm for extracting electrical parameters includes: Algorithms for extracting saturation current, off-state current, on / off ratio, threshold voltage, subthreshold swing, maximum transconductance, and hysteresis from a single transfer characteristic curve; algorithms for calculating drain-induced barrier reduction from multiple transfer characteristic curves; judging whether a device has failed based on the transfer characteristic curve, and the criteria for failure judgment include channel short circuit, channel open circuit, gate leakage, logic error, and subthreshold anomaly; algorithms for extracting threshold, gain, hysteresis, and noise margin from a single inverter voltage transfer curve; Multiple curves are extracted respectively, and the mean and standard deviation of the extracted multiple electrical parameters are calculated and analyzed.

4. The online data processing method according to claim 2, characterized in that: include: The customized conversion protocol is formed into source code containing different STEXs, so that it is compatible with the parameter extraction and analysis protocols of the following test devices: field effect transistors, inverters, resistor elements, and diodes; Embed the customized transformation protocol as a plug-in into the data analysis module.

5. The online data processing method according to claim 3, characterized in that: Further including: Through the data analysis module, the algorithm for extracting electrical parameters is called to extract, analyze, plot and generate reports for the data, and the obtained data and reports are sent to the client for visual display.

6. The online data processing method according to claim 2, characterized in that: include: The prediction model built based on the large language model refers to collecting user information data, the test equipment data used by the user, the image type generated by the preference, test items, test report format, and saving habits, and marking the preference items and report types for the above data, so that the prediction model can learn the relationship between the above data. When the user information data is input, the corresponding test report template can be automatically generated according to the information items preferred by the user.

7. The online data processing method according to claim 4, characterized in that: Further including: According to the device batches in the sample information, parse, extract, and draw in sequence to generate a plurality of different source code files for generating corresponding test reports; The sample information includes process name, device batch, test time, tester, and test instrument; Specifically parse the file elements in the Doc list and render them. When rendering, read each element in the Doc list in turn and render them into corresponding formats. The file elements include multi-level titles, texts corresponding to each level of titles, tables, and images.

8. The online data processing method according to claim 1, characterized in that: The step of extracting linear regions corresponding to different electrical parameters according to the characteristics of the target curve and the characteristics of the linear region to be extracted comprises: For carbon-based devices, when the target curve is the transfer characteristic curve of a field effect transistor, the linear regions corresponding to different electrical parameters are extracted including: Derivative the curve formed by the original test data and take the absolute value to find the reference point with the largest absolute value; Repeatedly collect data points on both sides of the reference point until the number of data points is less than or equal to the specified number and a data point less than the specified value is encountered, and use the collected data points as the linear region to further perform linear fitting; The electrical parameters include saturation current, off-state current, on / off ratio, transconductance, threshold voltage, drain-induced barrier lowering, and subthreshold swing.

9. An online data processing system, characterized in that: It executes the online data processing method according to any one of claims 1 to 8, and the online data processing system comprises: A receiving and processing module, used to receive an access request from the current client, and obtain original test data based on the access request; The network module is used to extract electrical parameters from the original test data and determine the image and sample information to be generated; the electrical parameters include electrical parameters related to the IV curve, the transfer characteristic curve and the inverter voltage transfer curve, specifically including: extracting linear regions corresponding to different electrical parameters according to the characteristics of the target curve and the characteristics of the linear region to be extracted; determining the reference value or reference range of each electrical parameter according to the extracted linear region; The data analysis module performs formatting processing using a customized conversion protocol based on the extracted electrical parameters, the determined images to be generated, and the sample information, and determines a customized template matching the current client based on the current client, so as to generate corresponding test reports in batches according to the determined customized template; The report generation module visualizes the extracted electrical parameters and the generated test report based on the input parameters of the current client.

10. The online data processing system according to claim 9, characterized in that: include: The network module can communicate with the storage module, the data analysis module, and the report generation module, and the data analysis module is integrated with an algorithm for extracting electrical parameters; Test reports include those related to the performance of field effect transistors, inverters, resistor elements, and diodes; The electrical parameters include saturation current, off-state current, on / off ratio, transconductance, threshold voltage, drain-induced barrier lowering, and subthreshold swing.

Images