Dram data retention time prediction method, apparatus, device, and medium

Abstract

The present disclosure provides a DRAM data retention time prediction method, device, equipment and medium, relating to the technical field of memory testing. The method comprises collecting to-be-fitted data under the DRAM data retention time in a first preset time range; fitting the to-be-fitted data through a preset lognormal distribution function to obtain key parameters, and in the case that the key parameters meet a preset fitting condition, obtaining a target lognormal distribution function; and determining the data retention time of the DRAM according to the target lognormal distribution function. The DRAM data retention time prediction method, device, equipment and medium provided by the present disclosure can avoid the influence of intrinsic failure, accurately predict the DRAM data retention time test, and quickly locate the cause of yield loss.

Description

Technical Field

[0001] This disclosure relates to the field of memory testing technology, and in particular to a DRAM data retention time prediction method, a DRAM data retention time prediction device, an electronic device, and a computer-readable storage medium. Background Technology

[0002] For volatile memories, such as DRAM (Dynamic Random Access Memory), the original stored data cannot be retained after the memory loses power. A DRAM memory cell typically includes a transistor and a capacitor. The retention time of a DRAM memory cell refers to the effective time during which the memory cell can maintain the accuracy of the stored data without causing data read failure after the capacitor leaks current.

[0003] The manufacturing process of memory has process deviations that cause the retention time of some memory cells to change as the testing time increases.

[0004] In related technologies, DRAM data retention time is currently typically characterized using WAT (Wafer Access Test) or CP (Circuit Probe). However, WAT cannot accurately reflect DRAM data storage time performance and has low test accuracy; while CP testing requires testing the failure rate of DRAM in the range of one to several hundred milliseconds, which is time-consuming and costly.

[0005] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] This disclosure provides a method, apparatus, device, and medium for predicting DRAM data retention time, which at least to some extent overcomes the problems of low accuracy and long testing time in DRAM data retention time prediction methods provided in related technologies.

[0007] Other features and advantages of this disclosure will become apparent from the following detailed description, or may be learned in part from practice of this disclosure.

[0008] According to one aspect of this disclosure, a method for predicting DRAM data retention time is provided, comprising: acquiring data to be fitted within a first preset time range of DRAM data retention time; fitting the data to be fitted using a preset log-normal distribution function to obtain key parameters; obtaining a target log-normal distribution function when the key parameters satisfy preset fitting conditions; and determining the DRAM data retention time based on the target log-normal distribution function.

[0009] In one embodiment of this disclosure, the key parameters include the mean, standard deviation, slope of the target log-normal distribution function within the first preset time range, and intercept of the target log-normal distribution function on the time axis.

[0010] In one embodiment of this disclosure, the key parameters satisfying the preset fitting conditions include: the standard deviation and the reciprocal of the slope are linearly related, and the goodness of fit between the standard deviation and the reciprocal of the slope is greater than or equal to a preset accuracy threshold.

[0011] In one embodiment of this disclosure, the goodness of fit includes at least one of the following: coefficient of determination R. 2 Root mean square error, nonlinear correlation coefficient R NL .

[0012] In one embodiment of this disclosure, determining the data retention time of the DRAM based on the target log-normal distribution function includes: obtaining the coordinate value on the time axis corresponding to the preset failure rate based on the preset failure rate and the target log-normal distribution function, and using it as the data retention time of the DRAM.

[0013] In one embodiment of this disclosure, the method further includes: when the key parameters do not meet the preset fitting conditions, collecting the data to be fitted under the DRAM data retention time of the second preset time range, refitting until the preset fitting conditions are met, and obtaining the target log-normal distribution function.

[0014] In one embodiment of this disclosure, the second preset time range is greater than the first preset time range.

[0015] In one embodiment of this disclosure, the method further includes: if the key parameter does not meet the preset fitting conditions, selecting a second preset time range based on the key parameter.

[0016] In one embodiment of this disclosure, the type of the data to be fitted includes at least one of the following: gate-drain leakage current, channel leakage current, junction current, and dielectric leakage current.

[0017] In one embodiment of this disclosure, the preset log-normal distribution function includes: a log-normal cumulative function; a log-normal probability density function; an exponential fitting function; a log-normal cumulative function and an exponential fitting function; or, a log-normal probability density function and an exponential fitting function.

[0018] In one embodiment of this disclosure, the step of collecting the data to be fitted under the DRAM data retention time within a first preset time range includes: testing the DRAM to obtain the DRAM shmoo test result; obtaining the failure rate of the DRAM data retention time within the first preset time range based on the shmoo test result; and using the failure rate as the data to be fitted to perform log-normal distribution fitting.

[0019] In one embodiment of this disclosure, the method further includes: determining process parameters for fabricating DRAM based on the key parameters.

[0020] According to another aspect of this disclosure, a DRAM data retention time prediction device is provided, comprising: a data acquisition module for acquiring data to be fitted within a first preset time range of DRAM data retention time; a data fitting module for fitting the data to be fitted using a preset log-normal distribution function to obtain key parameters, and obtaining a target log-normal distribution function when the key parameters satisfy preset fitting conditions; and a determination module for determining the DRAM data retention time based on the target log-normal distribution function.

[0021] According to another aspect of this disclosure, an electronic device, a processor, and a memory are provided, the memory being used to store executable instructions of the processor; wherein the processor is configured to perform the above-described DRAM data retention time prediction method by executing the executable instructions.

[0022] According to another aspect of this disclosure, a computer-readable storage medium is provided having a computer program stored thereon, characterized in that the computer program, when executed by a processor, implements the above-described DRAM data retention time prediction method.

[0023] According to another aspect of this disclosure, a computer program product is provided, the computer program product comprising a computer program or computer instructions, the computer program or computer instructions being loaded and executed by a processor to enable a computer to implement the above-described indicator market management method.

[0024] This disclosure provides a method, apparatus, device, and medium for predicting DRAM data retention time. The method involves collecting data to be fitted within a first preset time range of DRAM data retention time; fitting the data to be fitted using a preset log-normal distribution function to obtain key parameters; obtaining a target log-normal distribution function when the key parameters meet preset fitting conditions; and determining the DRAM data retention time based on the target log-normal distribution function. This avoids interference from numerous defects and intrinsic failures on DRAM data retention time evaluation, accurately predicts DRAM data retention time testing, significantly reduces the amount of data to be fitted, shortens sampling time, and quickly locates the causes of yield loss.

[0025] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0026] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure. It is obvious that the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0027] Figure 1 This diagram illustrates a DRAM data retention time prediction method according to an embodiment of the present disclosure.

[0028] Figure 2 This diagram illustrates a flowchart of a method for acquiring data to be fitted, as provided in an embodiment of this disclosure.

[0029] Figure 3 This diagram illustrates another DRAM data retention time prediction method provided by an embodiment of the present disclosure.

[0030] Figure 4 This diagram illustrates a flowchart of yet another DRAM data retention time prediction method provided in an embodiment of the present disclosure;

[0031] Figure 5 This diagram illustrates a flowchart of yet another DRAM data retention time prediction method provided in an embodiment of the present disclosure;

[0032] Figure 6 This diagram illustrates the theoretical average GIDL distribution provided in the embodiments of this disclosure.

[0033] Figure 7 Show Figure 6 A graph of the probability density function PDF;

[0034] Figure 8This illustration shows a fitting graph between σ and 1 / k, theoretically obtained based on fitting a log-normal distribution function, according to an embodiment of this disclosure.

[0035] Figure 9 This diagram illustrates the failure rate distribution of different products provided in embodiments of this disclosure over different time periods;

[0036] Figure 10 This diagram illustrates the influence of the standard deviation σ, a key parameter of the log-normal distribution function, on the fitting results according to an embodiment of this disclosure.

[0037] Figure 11 This diagram illustrates the impact of the WAT method on data retention time in related technologies.

[0038] Figure 12 This diagram illustrates the influence of the intercept b, a key parameter of the log-normal distribution function, on the fitting results according to an embodiment of the present disclosure.

[0039] Figure 13 This diagram shows the CDF fitting curve at 1E-7 according to an embodiment of the present disclosure;

[0040] Figure 14 This diagram illustrates the goodness-of-fit at 1E-7 according to an embodiment of the present disclosure.

[0041] Figure 15 This diagram shows the CDF fitting curve at 1E-5 according to an embodiment of the present disclosure;

[0042] Figure 16 This diagram illustrates the goodness-of-fit at 1E-5 according to an embodiment of the present disclosure.

[0043] Figure 17 This diagram shows a comparison of the fitting results of Ave FIT and P50 FIT according to embodiments of this disclosure;

[0044] Figure 18 This invention discloses a statistical graph showing the standard deviation data obtained by fitting different products according to embodiments of the present invention.

[0045] Figure 19 This diagram illustrates the statistical graph of intercept data obtained by fitting different products according to embodiments of this disclosure;

[0046] Figure 20 This diagram shows a statistical graph of goodness-of-fit data from embodiments of this disclosure.

[0047] Figure 21 This diagram illustrates a DRAM data retention time prediction device according to an embodiment of the present disclosure.

[0048] Figure 22 This diagram illustrates the structure of an electronic device provided in an embodiment of the present disclosure. Detailed Implementation

[0049] Preferred embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While preferred embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0050] The terms "first" and "second" used in this document are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly stated.

[0051] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection, an electrical connection, or a connection that allows communication between them; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication between two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0052] The following disclosure provides many different implementations or examples for carrying out different structures of this application. To simplify the disclosure, specific examples of components and arrangements are described below. Of course, these are merely examples and are not intended to limit the scope of this application. Furthermore, reference numerals and / or reference letters may be repeated in different examples; such repetition is for simplification and clarity and does not in itself indicate a relationship between the various implementations and / or arrangements discussed.

[0053] For DRAM data retention time, WAT or CP are commonly used in related technologies to characterize it; however, both testing methods have certain limitations.

[0054] The WAT testing method can only test multiple DRAM memory structures and calculate the average value as the DRAM data retention time. However, since the DRAM data retention time is mainly limited by certain types of cells with high leakage current, the average value cannot truly reflect the DRAM's data storage time performance.

[0055] CP testing can measure the failure rate of DRAM over a wide time range (e.g., from one millisecond to several hundred milliseconds), thereby evaluating the data retention time performance of DRAM based on the failure rate. However, in the early stages of R&D, products have a large number of defects and intrinsic failures. These defects and intrinsic failures will interfere with the evaluation of DRAM data retention time. In addition, CP testing requires collecting DRAM failure rate data over a wide time range, resulting in long sampling times and high testing costs, making it impossible to provide accurate and reliable evaluation data for R&D or production in a timely manner.

[0056] There are various mechanisms affecting DRAM leakage, such as dielectric leakage, PN junction leakage, channel leakage, gate leakage, etc. These leakage mechanisms can be broadly classified into two categories: dielectric leakage and transistor leakage. Both types of leakage follow a log-normal distribution. Based on this, the DRAM data retention time prediction method provided in this disclosure collects data to be fitted within a first preset time range of DRAM data retention time; fits the data to be fitted using a preset log-normal distribution function to obtain key parameters; when the key parameters meet preset fitting conditions, a target log-normal distribution function is obtained; and determines the DRAM data retention time based on the target log-normal distribution function. This method utilizes the log-normal distribution to accurately analyze and predict DRAM data retention time, avoiding interference from numerous defects and intrinsic failures on DRAM data retention time evaluation, accurately predicting DRAM data retention time testing, significantly reducing the amount of data to be fitted, shortening sampling time, and quickly locating the causes of yield loss.

[0057] It should be noted that, unless otherwise specified, the embodiments of the present invention and the technical features thereof can be combined with each other.

[0058] The following detailed description of this exemplary implementation method is provided in conjunction with the accompanying drawings and embodiments.

[0059] Figure 1 A flowchart illustrating a DRAM data retention time prediction method provided in an embodiment of this disclosure is shown. Figure 1 As shown, the DRAM data retention time prediction method provided in this embodiment includes:

[0060] S102. Collect the data to be fitted within the DRAM data retention time of the first preset time range.

[0061] The preset time range in this embodiment is a specific time range selected within the time range of DRAM leakage current testing, and the first preset time range is one of the specific time ranges.

[0062] For example, the time range for DRAM leakage testing can be divided into multiple intervals. The right endpoint of the time range can be determined according to the actual situation, such as (0, 100ms], (0, 200ms], (0, 300ms], etc.

[0063] The first preset time range can be selected as either a time range with a smaller right endpoint or a time range with a larger right endpoint.

[0064] The aforementioned data to be fitted is obtained by processing DRAM leakage test data. For example, by performing a leakage test on DRAM, the DRAM shmoo test results can be obtained. The shmoo test results are then processed to obtain the failure rate of DRAM data retention time within a first preset time range, which is used as the data to be fitted.

[0065] In one embodiment, the type of data to be fitted includes at least one of the following: gate-induced drain leakage (GIDL), channel leakage current, junction current, and dielectric leakage current. Various factors can affect DRAM leakage, such as defect quantity, trap energy level, and electric field. Taking gate-induced drain leakage (GIDL) as an example, the causes of GIDL may include tunneling probability, subthreshold leakage, and junction leakage.

[0066] It should be noted that, in addition to using the Shmoo test results to characterize leakage current data, other methods that can characterize leakage current can also be used, and this disclosure does not make any specific limitations.

[0067] S104. Fit the data to be fitted using a preset log-normal distribution function to obtain key parameters. If the key parameters meet the preset fitting conditions, obtain the target log-normal distribution function.

[0068] In one embodiment, the preset log-normal distribution function includes: a log-normal distribution cumulative distribution function (CDF); a log-normal distribution probability density function (PDF); an exponential fitting function; a log-normal distribution cumulative distribution function and an exponential fitting function; or, a log-normal distribution probability density function and an exponential fitting function, etc. This disclosure, based on the distribution pattern of leakage current types in DRAM, presets multiple log-normal distribution functions to fit the data to be fitted, thereby improving the prediction accuracy of DRAM data retention time, shortening measurement time, and reducing the amount of data to be fitted.

[0069] The cumulative function of a log-normal distribution with standard deviation σ and mean μ is:

[0070]

[0071] in, Let x be the log-normal probability density function, x be the independent variable, lnx be the logarithm to the base e, and erf be the error function.

[0072] The error function erf with independent variable x is defined as:

[0073]

[0074] Where, erf(∞)=1 and erf(-x)=-erf(x).

[0075] The complementary error function erfc(x) with independent variable x is defined as:

[0076]

[0077] The probability density function of a log-normal distribution with standard deviation σ and mean μ can be obtained by taking the derivative of the cumulative function of the log-normal distribution. Alternatively, the probability density function of a log-normal distribution can also be obtained by replacing the independent variable in a normal distribution.

[0078] Since Z follows a standard normal distribution, μ + Zσ follows a normal distribution with mean μ and standard deviation σ. Because lnX = μ + Zσ, lnX also follows a normal distribution with mean μ and standard deviation σ, i.e.:

[0079]

[0080] By converting logarithmic coordinates to linear coordinates and labeling X as x, the probability density function of the log-normal distribution of x in linear coordinates is obtained as follows:

[0081]

[0082] It should be noted that the above-mentioned log-normal cumulative function and log-normal probability density function are based on the natural logarithm e. In actual implementation, scientific notation 10 or other values ​​can also be used as the base, requiring only simple transformations. This disclosure does not impose any specific limitations.

[0083] In a log-log coordinate system, the tail of the cumulative function CDF or the probability density function PDF (i.e., the portion of the curve near the origin of the coordinate system, or the portion of the curve over a smaller time range as disclosed herein) is linear. Based on this, the exponential fitting function can be expressed as:

[0084] lny = k(lnx - b) or,

[0085] y = e k(lnx-b) =x k ×e -kb (Formula Six)

[0086] Where y is the ordinate of the cumulative function CDF or probability density function PDF, x is the abscissa of the cumulative function CDF or probability density function PDF, k is the slope of the linear line segment, and b is the intercept of the linear line segment on the abscissa of the coordinate system (i.e., the time axis of this disclosure).

[0087] At this point, the expected value E(x) and variance D(x) of the log-normal distribution are respectively:

[0088]

[0089]

[0090] It should be noted that key parameters may include the mean, standard deviation, slope of the target log-normal distribution function within the first preset time range, and intercept of the target log-normal distribution function on the time axis.

[0091] This disclosure utilizes Python tools to perform parameter fitting under a log-normal distribution condition, thereby obtaining the aforementioned key parameters.

[0092] When fitting data to be fitted using Python tools, the CDF (Formula 1), PDF (Formula 5), ​​and exponential fitting function (Formula 6) mentioned above can be rewritten.

[0093] CDF (Formula 5) can be rewritten as:

[0094] y1=np.exp(-(np.log(x)-u)**2 / (2*sig**2)) / (np.sqrt(2*np.pi)*sig*x);

[0095] PDF (Formula 1) rewritten as:

[0096] y2=0.5*(1+math.erf((np.log(x)-u) / (np.sqrt(2)*sig)));

[0097] The exponential fitting function (Formula 6) can be rewritten as follows:

[0098] y3 = pow(10,b) * pow(x,k).

[0099] The aforementioned preset fitting conditions are used to characterize the degree of matching between the log-normal distribution function and the data to be fitted. They can also be understood as the minimum standard or minimum requirement for the log-normal distribution function to be fitted to the data to be fitted.

[0100] When the key parameters meet the preset fitting conditions, fitting the data to be fitted using the target log-normal distribution function is as expected, and the target log-normal distribution function is obtained. The target log-normal distribution function is a log-normal distribution function that can match the distribution pattern of the data to be fitted to a good degree. When the key parameters do not meet the preset fitting conditions, it indicates that the result of fitting the data to be fitted using the preset log-normal distribution function cannot achieve the expected result. This indicates that the amount of data collected is too small, and it is necessary to collect more data to be fitted than the previous fitting.

[0101] It should be noted that the preset fitting conditions can be pre-configured in the data fitting module of the test equipment. The specific value range and type of the preset fitting conditions can be determined according to the specific situation.

[0102] S106. Determine the data retention time of the DRAM based on the target log-normal distribution function.

[0103] In one embodiment, given the target log-normal distribution function, the DRAM data retention time can be determined based on the target log-normal distribution function. Specifically, based on the preset failure rate (i.e., the preset coordinate value of the vertical axis of the normal distribution function) and the target log-normal distribution function, the coordinate value on the time axis corresponding to the preset failure rate is obtained and used as the DRAM data retention time. That is, the preset failure rate is substituted into the target log-normal distribution function, and the coordinate value of the time axis obtained is the DRAM data retention time.

[0104] The DRAM data retention time prediction method provided in this disclosure collects data to be fitted within a first preset time range of DRAM data retention time; fits the data to be fitted using a preset log-normal distribution function to obtain key parameters; when the key parameters meet preset fitting conditions, a target log-normal distribution function is obtained; and determines the DRAM data retention time based on the target log-normal distribution function. This avoids interference from a large number of defects and intrinsic failures on the DRAM data retention time evaluation, accurately predicts DRAM data retention time testing, significantly reduces the amount of data to be fitted, shortens sampling time, and quickly locates the cause of yield loss.

[0105] Figure 2 A flowchart illustrating the method for acquiring data to be fitted according to an embodiment of this disclosure is shown. Figure 2 In one embodiment, the above-mentioned S102, which involves collecting the data to be fitted within a first preset time range of DRAM data retention time, includes:

[0106] S1022. Test the DRAM and obtain the DRAM shmoo test results;

[0107] S1022. Based on the shmoo test results, obtain the failure rate of DRAM data retention time within the first preset time range, and use the failure rate as the data to be fitted to perform log-normal distribution fitting.

[0108] The DRAM SHMO measurement results are used to characterize the state of different DRAM cells by using the color of a graphical block, with different data retention times on the horizontal axis and DRAM leakage current on the vertical axis. The color of the graphical block can include red and green, where red indicates that the DRAM cell is in a failed state and green indicates that the DRAM cell is in a normal state. When the leakage current of the DRAM cell is greater than a preset leakage current threshold, the DRAM cell is determined to be in a failed state; when the leakage current of the DRAM cell is less than or equal to the preset leakage current threshold, the DRAM cell is determined to be in a normal state. The preset leakage current threshold is determined according to the actual situation and is not specifically limited in this disclosure.

[0109] The failure rate within the first preset time range of DRAM data retention time is the ratio between the number of graphics blocks displaying failure states (the number of red graphics blocks in the Shmoo test results) and the total number of graphics blocks (the sum of the number of red and green graphics blocks in the Shmoo test results). For example... Figure 9 As shown, a schematic diagram of the failure rate distribution of different products within different time ranges is provided, where the horizontal axis represents the pause time and the vertical axis represents the failure rate (Cumulative Probility) corresponding to different pause times.

[0110] The shmoo test results can collect test results from different batches of products, different substrates, and different test time ranges or distribution ranges.

[0111] Based on the shmmo test results of DRAM, this disclosure statistically analyzes the failure rate of DRAM data retention time within a first time range as the data to be fitted, thereby achieving log-normal distribution fitting. Different types and quantities of data to be fitted ensure the accuracy of the log-normal distribution in predicting DRAM data retention time.

[0112] Figure 3 A flowchart illustrating another DRAM data retention time prediction method provided in an embodiment of this disclosure is shown. Figure 1 Based on the example, the key parameters in S104 that satisfy the preset fitting conditions are further refined into S1042 to S1044 to determine the scheme in which the key parameters satisfy the preset fitting conditions. For example... Figure 3 As shown, the DRAM data retention time prediction method provided in this embodiment includes S102, S1042-S1044, and S106. Specifically, the method includes:

[0113] S1042. Fit the data to be fitted using a preset log-normal distribution function to obtain key parameters;

[0114] S1044, The standard deviation and the reciprocal of the slope are linearly related, and the goodness of fit between the standard deviation and the reciprocal of the slope is greater than or equal to the preset accuracy threshold.

[0115] It should be noted that the implementation methods of S102, S1042, and S106 are the same as the specific implementation methods of the above embodiments, and will not be repeated here.

[0116] In this embodiment, key parameters include the standard deviation σ and the slope k of the target log-normal distribution function within a first preset time range. As can be seen from the above embodiment, the target log-normal distribution function is linear at the tail of the curve, and this linear relationship is defined by the standard deviation and the slope, i.e., σ ~ 1 / k exhibits a linear relationship.

[0117] To characterize the degree to which the standard deviation σ and the reciprocal of the slope k are linear, this disclosure determines the relationship by the goodness of fit between the two. When the goodness of fit between the standard deviation and the reciprocal of the slope exceeds a preset accuracy threshold, the two are determined to be linearly related; when the goodness of fit between the standard deviation and the reciprocal of the slope does not exceed the preset accuracy threshold, the two are determined to be nonlinearly related.

[0118] In this disclosure, goodness of fit includes at least one of the following: coefficient of determination R. 2 Root mean square error, nonlinear correlation coefficient R NL .

[0119] Root mean square error (RMSE) is the expected value of the squared difference between the predicted and actual values ​​of a parameter. MSE can evaluate the degree of variation in data; the smaller the MSE value, the better the accuracy of the prediction model in describing the experimental data. The MSE is expressed by the following formula:

[0120]

[0121] Among them, y i For the actual value of the parameter, Here are the predicted values ​​for the parameters, and n is the total number of parameters.

[0122] The coefficient of determination compares the predicted value to the value obtained using only the mean, determining the amount of improvement. The coefficient of determination ranges from 0 to 1; a better model fit results in a coefficient of determination closer to 1, and a worse fit results in a coefficient of determination closer to 0. The coefficient of determination R0 2 It can be expressed as follows:

[0123]

[0124] Among them, y i For the actual value of the parameter, Here are the predicted values ​​for the parameters, and n is the total number of parameters. This represents the average value of the parameters.

[0125] R NL The closer the value is to 1, the better the model fit and the higher the fitting accuracy R. NL We obtain it from the following formula:

[0126]

[0127] Where y is the actual value of the parameter. y i For the actual value of the parameter, These are the predicted values ​​for the parameters.

[0128] Depending on the type of goodness of fit, different values ​​can be set for the preset accuracy threshold, which can be pre-configured in the data fitting module. For example, the coefficient of determination R can be set to a different value. 2 or R NL The preset accuracy thresholds are set to 0.8, 0.9, 0.99, etc.

[0129] It should be noted that the above-mentioned preset accuracy threshold is only a specific example provided to illustrate the embodiments of this disclosure, and should not be regarded as a limitation on the scope of protection of this disclosure. In the process of fitting the data to be fitted by the log-normal distribution function for the DRAM data retention time, the value of the preset accuracy threshold can also be determined according to the specific situation. This disclosure does not make any specific limitation.

[0130] This disclosure utilizes the linear relationship between the standard deviation and the reciprocal of the slope at the tail of the log-normal distribution function curve to fit a log-normal distribution, and further utilizes the coefficients of determination, root mean square error, and R² to perform fitting. NL The goodness-of-fit is used to confirm the matching degree of the above linear relationship, thereby achieving accurate prediction of the data to be fitted within a small preset time range and ensuring the accuracy of model fitting.

[0131] Figure 4 This diagram illustrates a flowchart of yet another DRAM data retention time prediction method provided in an embodiment of this disclosure. Figure 1 Based on the example, step S104 is further refined into S1042 and S1046 to determine how to determine the target log-normal distribution function when the key parameters do not meet the preset fitting conditions. Figure 4 As shown, in one embodiment, the prediction method includes steps S102, S1042, S1046, and S106. Specifically, the method includes:

[0132] S1046. If the key parameters do not meet the preset fitting conditions, collect the data to be fitted within the DRAM data holding time of the second preset time range, and refit until the preset fitting conditions are met to obtain the target log-normal distribution function.

[0133] It should be noted that the implementation methods of S102, S1042, and S106 are the same as the specific implementation methods of the aforementioned embodiments, and will not be repeated here.

[0134] The second preset time range is larger than the first preset time range. For example, the first preset time range can be a failure rate within 100ms, and the second preset time range can be a failure rate within 200ms. When the key parameters obtained by fitting a log-normal distribution to the failure rate within 100ms do not meet the preset fitting conditions, the data to be fitted under the DRAM data retention time within 200ms is re-acquired. The re-acquired data to be fitted is then fitted with a log-normal distribution, and the key parameters are judged again to see if they meet the preset fitting conditions. If they meet the preset fitting conditions, the target log-normal distribution function is obtained. Then, the DRAM data retention time is determined based on the target log-normal distribution function. Starting from a smaller time range or a smaller number of data to be fitted, the time range or the number of data to be fitted is gradually expanded to determine the target log-normal distribution function. This can improve the prediction efficiency of DRAM data retention time and shorten the prediction time.

[0135] In one embodiment, if the key parameters do not meet the preset fitting conditions, a second preset time range is selected based on the key parameters.

[0136] When the key parameters do not meet the preset fitting conditions, the coefficients of determination R obtained from the fitting can also be used. 2 The value of and the first preset time range are used to determine the second preset time range. For example, if the first preset time range is 100ms, the coefficient R is determined. 2 The coefficient of determination is 0.237, indicating that the data to be fitted within the first preset time range of DRAM data retention time exhibits poor linearity at the tail of the log-normal distribution function curve. A larger amount of data needs to be selected to determine the subsequent curve's pattern. Therefore, the second preset time range can be set to 300ms. If the first preset time range is 100ms, the coefficient of determination R... 2 The value of 0.85 indicates that the linearity of the data to be fitted under the DRAM data retention time in the first preset time range is poor at the tail of the log-normal distribution function curve. By increasing a small amount of data to be fitted, the pattern of the subsequent curve can be determined. Therefore, the second preset time range can be selected as 200ms.

[0137] This disclosure determines a second preset time range based on the relationship between the key parameters and the preset fitting conditions, as well as a first preset time range, when the key parameters do not meet the preset fitting conditions, thereby reducing the number of fitting attempts and improving prediction efficiency.

[0138] Figure 5 This diagram illustrates a flowchart of yet another DRAM data retention time prediction method provided in an embodiment of this disclosure. Figure 1 Based on the previous example, step S108 is added after step S104 to determine the process parameters for subsequent DRAM fabrication based on the current fitting results. Figure 5 As shown, the DRAM data retention time prediction method provided in the embodiment includes S102 to S108. Specifically, the method includes:

[0139] S108. Determine the process parameters for DRAM fabrication based on the key parameters.

[0140] It should be noted that the implementation methods of S102 to S106 are the same as the specific implementation methods provided in the previous embodiments, and will not be repeated here.

[0141] In a log-normal distribution function, the mean μ and standard deviation σ, two key parameters, determine the shape and position of the curve. The mean determines the main distribution of the log-normal distribution, not the tail portion of the curve; while the standard deviation σ determines the width of the curve. Typically, as the leakage current distribution gradually worsens, the standard deviation σ increases, and even if the mean μ is small, it still has a significant impact on the shape of the curve.

[0142] In some cases, the slope of one log-normal distribution curve is greater than that of another log-normal distribution curve. It is also possible that a log-normal distribution curve with a larger slope is worse at the tail than a log-normal distribution curve with a smaller slope.

[0143] In one embodiment, based on the magnitude of the mean μ and standard deviation σ of the key parameters, it can be determined whether the process parameters for DRAM fabrication can meet the requirements. When the process parameters cannot meet the requirements, the cause of leakage is analyzed in conjunction with the specific type of the data to be fitted, and then the process parameters for DRAM fabrication are adjusted to reduce leakage, so that the performance distribution of the adjusted DRAM is more concentrated (the standard deviation σ is reduced) and the mean is closer to the required value.

[0144] This disclosure determines the process parameters for DRAM fabrication through key parameters, which can provide effective data retention time evaluation parameters during DRAM product development, guide the direction of process optimization, help shorten development time, and provide accurate and reliable data retention time prediction data for development.

[0145] To gain a deeper understanding of the implementation of the embodiments of this disclosure, the following uses a tested GIDL as an example, in conjunction with the appendix. Figure 6 ~Attached Figure 19 Please provide a detailed explanation.

[0146] Figure 6 This diagram illustrates the relationship between the theoretically average GIDL and the average gate-drain overlap length (Lov). The average GIDL and average Lov are measured using a Test Element Group (TEG). The average Lov is calculated indirectly using the overlap capacitance between the gate and drain. The average GIDL increases exponentially with the average Lov. Both follow a log-normal distribution. Based on Equation 1, we can obtain... Figure 6 The curve of the probability density function PDF, as shown in the figure. Figure 7 As shown, after fitting with the log-normal distribution probability density function PDF, σ and 1 / k have a linear relationship, and the coefficient of determination R0 2 =0.99921, indicating that the following was adopted. Figure 7 PDF curve implementation for Figure 6 GIDL fits well with Lov.

[0147] In this disclosure, measurements were performed on different wafers (AP02, AP11, AP21, AP35) of the AP01 batch of DRAM to obtain GIDL Shmoo images for different wafers. Based on the GIDL Shmoo images, the number of graphic blocks displaying failure states and the sum of the number of image block failure states and the number of normal states were counted within different time ranges. The ratio of the two was calculated to obtain the failure rate (e.g., ...) within different time ranges. Figure 9 (as shown), Figure 9 The straight lines in the figure represent the trend of failure rate over different time ranges, and the failure rate is approximately linearly related to the pause time (the dashed lines in the figure). As shown in Table 1, the failure rates obtained by fitting the average value and the median value for different batches of wafers are presented respectively.

[0148] Table 1 Failure rates of different batches of wafers

[0149] batch Wafer ID AVG P50 AP01 #2 2.46E-05 1.97E-05 AP01 #11 5.36E-06 5.54E-06 AP01 #21 3.01E-05 2.89E-05 AP01 #25 1.93E-04 1.06E-04

[0150] In Table 1, AVG represents the failure rate obtained by fitting the mean value, and P50 represents the failure rate obtained by fitting the median value.

[0151] Given the mean μ and standard deviation σ, by Figure 10 It can be observed that as the standard deviation σ increases, the tail of the log-normal distribution curve rises, and the mean μ determines the main position of the log-normal distribution curve.

[0152] For the intercept of the log-normal distribution function on the time axis, from Figure 11 The fitting results of AVEFIT using the mean value show a significant rise at the tail of the data. Figure 12 The fitting results of the P50 FIT data fitted using the median show a good match.

[0153] For the log-normal distribution function fitting results of CDF in the range of 1E-7 and the corresponding linear relationship between σ and 1 / k, as follows: Figure 13 and Figure 14 As shown in the figure. It can be observed from the figure that the coefficient of determination R obtained by fitting within the range of 1E-7 is... 2 =0.237. When the preset accuracy threshold is 0.8, it indicates that the key parameters of the preset time range cannot meet the preset fitting conditions, and the preset time range needs to be reconfirmed.

[0154] For the log-normal distribution function fitting results of CDF in the range of 1E-5 and the corresponding linear relationship between σ and 1 / k, as follows: Figure 15 and Figure 16 As shown in the figure. It can be observed from the figure that the coefficient of determination R obtained by fitting within the range of 1E-5 is...2 =0.845. When the preset accuracy threshold is 0.8, it indicates that the key parameters of the preset time range meet the preset fitting conditions, and the target log-normal distribution function is obtained.

[0155] Figure 17 This diagram shows a comparison of the fitting results of the mean-fitted AVE FIT and the median-fitted P50 FIT according to embodiments of this disclosure. Figure 17 It can be observed that, compared with fitting the data to be fitted using the mean value to the AVE FIT, fitting the data to be fitted using the median value to the P50 FIT achieves a better fit across the entire time range.

[0156] In this disclosure, the median is used to fit the P50 FIT data to be fitted, and a log-normal distribution function is used for fitting. The key parameter, standard deviation σ, is obtained (e.g., Figure 18 As shown), the intercept b of the target log-normal distribution function on the time axis (e.g.) Figure 19 (as shown), and the coefficient of determination R 2 and R NL (like Figure 20 (As shown). Where σ-1, σ-2, and σ-3 are the standard deviations obtained by fitting the P50 to different product generations; b-1, b-2, and b-3 are the intercepts obtained by fitting the P50 to different products. Figure 20 In this context, R²_log represents the coefficient of determination obtained by fitting in logarithmic coordinates; R²_lin represents the coefficient of determination obtained by fitting in linear coordinates; and RNL_log represents the nonlinear correlation coefficient R obtained by fitting in logarithmic coordinates. NL RNL_lin is the nonlinear correlation coefficient R obtained by fitting under linear coordinates. NL It should be noted that the fitting process for other data to be fitted (such as P05, P95, etc.) is similar to that of fitting the median to the P50 FIT, and will not be described in detail here.

[0157] After the above fitting, the key parameters and the corresponding data retention time prediction results can be obtained, as shown in Table 2.

[0158] Table 2. Results of fitting the log-normal distribution

[0159] X Y μ σ k b pre(1e-7) pre(1e-8) #2 #2 11.42847 1.144573 4.0139 -16.418 222.05 125.12 #11 #11 11.48375 1.153557 3.9951 -16.394 224.53 126.18 #21 #21 11.47301 1.151068 4.0034 -16.417 225.04 126.61 #25 #25 11.34922 1.159524 3.8648 -15.758 184.52 101.69

[0160] In Table 2, pre(1e-7) represents the selection of data to be fitted within the time range of (0, 1e-7) (i.e., the preset failure rate is 1e-7) to obtain the corresponding normal distribution function curve. Substituting y = 1e-7 into the above normal distribution function curve, the data retention time can be obtained. Similarly, pre(1e-8) represents the selection of data to be fitted within the time range of (0, 1e-8) (i.e., the preset failure rate is 1e-8) to obtain the corresponding normal distribution curve. Substituting y = 1e-8 into the above normal distribution function curve, the data retention time can be obtained.

[0161] In addition, the process parameters for subsequent DRAM fabrication can be optimized based on the aforementioned key parameters.

[0162] Based on the same inventive concept, this disclosure also provides a DRAM data retention time prediction device, as described in the following embodiments. Since the principle by which this device embodiment solves the problem is similar to that of the above-described method embodiment, the implementation of this device embodiment can refer to the implementation of the above-described method embodiment, and repeated details will not be elaborated further.

[0163] Figure 21 A schematic diagram of a DRAM data retention time prediction device according to an embodiment of the present disclosure is shown. Figure 21 As shown, the DRAM data retention time prediction device of this embodiment includes a data acquisition module 2101, a data fitting module 2102, and a determination module 2103.

[0164] Among them, the data acquisition module 2101 is used to acquire the data to be fitted under the DRAM data retention time within the first preset time range;

[0165] The data fitting module 2102 is used to fit the data to be fitted using a preset log-normal distribution function to obtain key parameters. Under the condition that the key parameters meet the preset fitting conditions, the target log-normal distribution function is obtained.

[0166] The determination module 2103 is used to determine the data retention time of the DRAM based on the target log-normal distribution function.

[0167] It should be noted that the key parameters include the mean, standard deviation, slope of the target log-normal distribution function within the first preset time range, and intercept of the target log-normal distribution function on the time axis.

[0168] In one embodiment of this disclosure, the data fitting module 2102 is used to determine whether the key parameters meet the preset fitting conditions. Specifically, the key parameters meeting the preset fitting conditions include: the standard deviation and the reciprocal of the slope are linearly related, and the goodness of fit between the standard deviation and the reciprocal of the slope is greater than or equal to a preset accuracy threshold.

[0169] It should be noted that goodness of fit includes at least one of the following: coefficient of determination R. 2 Root mean square error, nonlinear correlation coefficient R NL .

[0170] In one embodiment of this disclosure, the determining module 2103 is used to obtain the coordinate value on the time axis corresponding to the preset failure rate based on the preset failure rate and the target log-normal distribution function, and use it as the data retention time of the DRAM.

[0171] In one embodiment of this disclosure, the data fitting module 2102 is further configured to collect the data to be fitted within the DRAM data retention time of the second preset time range when the key parameters do not meet the preset fitting conditions, and refit until the preset fitting conditions are met to obtain the target log-normal distribution function.

[0172] It should be noted that the second preset time range is greater than the first preset time range.

[0173] In one embodiment, the data fitting module 2102 is further configured to select a second preset time range based on the key parameters if the key parameters do not meet the preset fitting conditions.

[0174] It should be noted that the data to be fitted includes at least one of the following: gate-drain leakage current, channel leakage current, junction current, and dielectric leakage current.

[0175] It should be noted that the preset log-normal distribution function includes: the log-normal cumulative function; the log-normal probability density function; the exponential fitting function; the log-normal cumulative function and the exponential fitting function; or the log-normal probability density function and the exponential fitting function.

[0176] In one embodiment of this disclosure, the data acquisition module 2101 is further configured to test the DRAM and obtain the DRAM shmoo test results; based on the shmoo test results, obtain the failure rate of the DRAM data retention time within a first preset time range, and use the failure rate as the data to be fitted to perform log-normal distribution fitting.

[0177] In one embodiment, the determining module 2103 is further configured to determine the process parameters for fabricating DRAM based on key parameters.

[0178] The DRAM data retention time prediction device provided in this embodiment collects data to be fitted within a first preset time range of DRAM data retention time; fits the data to be fitted using a preset log-normal distribution function to obtain key parameters; when the key parameters meet preset fitting conditions, a target log-normal distribution function is obtained; and determines the DRAM data retention time based on the target log-normal distribution function to avoid interference from a large number of defects and intrinsic failures on the DRAM data retention time evaluation, accurately predict DRAM data retention time testing, greatly reduce the amount of data to be fitted, shorten sampling time, and quickly locate the cause of yield loss.

[0179] Those skilled in the art will understand that various aspects of the present invention can be implemented as systems, methods, or program products. Therefore, various aspects of the present invention can be specifically implemented in the following forms: entirely hardware implementations, entirely software implementations (including firmware, microcode, etc.), or implementations combining hardware and software aspects, collectively referred to herein as “circuits,” “modules,” or “systems.”

[0180] The following reference Figure 22 To describe an electronic device 2200 according to this embodiment of the present invention. Figure 22 The electronic device 2200 shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0181] like Figure 22 As shown, the electronic device 2200 is manifested in the form of a general-purpose computing device. The components of the electronic device 2200 may include, but are not limited to: at least one processing unit 2210, at least one storage unit 2220, and a bus 2230 connecting different system components (including storage unit 2220 and processing unit 2210).

[0182] The storage unit stores program code that can be executed by the processing unit 2210, causing the processing unit 2210 to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of the present invention. For example, the processing unit 2210 can perform, as follows: Figure 1 The data to be fitted is collected within a first preset time range of DRAM data retention time. The data to be fitted is fitted using a preset log-normal distribution function to obtain key parameters. If the key parameters meet the preset fitting conditions, the target log-normal distribution function is obtained. The DRAM data retention time is determined based on the target log-normal distribution function.

[0183] Storage unit 2220 may include a readable medium in the form of a volatile storage unit, such as random access memory (RAM) 22201 and / or cache memory 22202, and may further include a read-only memory (ROM) 22203.

[0184] Storage unit 2220 may also include a program / utility 22204 having a set (at least one) program module 22205, such program module 22205 including but not limited to: operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0185] Bus 2230 can represent one or more of several types of bus structures, including memory cell bus or memory cell controller, peripheral bus, graphics acceleration port, processing unit, or local bus using any of the multiple bus structures.

[0186] Electronic device 2200 can also communicate with one or more external devices 2240 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with the system 2200, and / or with any device that enables the electronic device 2200 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 2250. Furthermore, the system can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 2260. Figure 22 As shown, network adapter 2260 communicates with other modules of electronic device 2200 via bus 2230. It should be understood that, although... Figure 22 As not shown, other hardware and / or software modules can be used in conjunction with electronic device 2200, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0187] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, terminal device, or network device, etc.) to execute the methods according to the embodiments of this disclosure.

[0188] In exemplary embodiments of this disclosure, a computer-readable storage medium is also provided, which may be a readable signal medium or a readable storage medium. In exemplary embodiments of this disclosure, a computer program product is also provided, comprising a computer program or computer instructions, which are loaded and executed by a processor to cause a computer to implement the steps of the methods disclosed in the above embodiments.

[0189] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0190] Furthermore, although the steps of the method in this disclosure are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or a step may be broken down into multiple steps.

[0191] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, mobile terminal, or network device, etc.) to execute the methods according to the embodiments of this disclosure.

[0192] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the appended claims.

Claims

1. A DRAM data retention time prediction method, characterized by, The method comprises: collecting to-be-fitted data of the DRAM in a first preset time range of data retention time; fitting the to-be-fitted data by using a preset lognormal distribution function to obtain key parameters, and obtaining a target lognormal distribution function in a case where the key parameters meet a preset fitting condition; determining the data retention time of the DRAM according to the target lognormal distribution function; the key parameters comprise a mean value, a standard deviation, a slope of the target lognormal distribution function in the first preset time range, and an intercept of the target lognormal distribution function on a time coordinate axis; the key parameters meeting the preset fitting condition comprises: the standard deviation and the reciprocal of the slope are in a linear relationship, and a fitting goodness of the standard deviation and the reciprocal of the slope is greater than or equal to a preset precision threshold.

2. The DRAM data retention time prediction method of claim 1, wherein, The fitting goodness comprises at least one of a determination coefficient R^2, a root mean square error, and a nonlinear correlation coefficient R_NL.

3. The DRAM data retention time prediction method of claim 1, wherein, The determining the data retention time of the DRAM according to the target lognormal distribution function comprises: obtaining a coordinate value corresponding to a preset failure rate on the time coordinate axis as the data retention time of the DRAM according to the preset failure rate and the target lognormal distribution function.

4. The DRAM data retention time prediction method of claim 1, wherein, The method further comprises: in a case where the key parameters do not meet the preset fitting condition, collecting to-be-fitted data of the DRAM in a second preset time range of data retention time, re-fitting until the preset fitting condition is met, and obtaining the target lognormal distribution function.

5. The DRAM data retention time prediction method of claim 4, wherein, The second preset time range is greater than the first preset time range.

6. The DRAM data retention time prediction method of claim 5, wherein, The method further comprises: in a case where the key parameters do not meet the preset fitting condition, selecting a second preset time range according to the key parameters.

7. The DRAM data retention time prediction method of claim 1, wherein, The type of the to-be-fitted data comprises at least one of a gate-induced drain leakage current, a channel leakage current, a junction current, and a dielectric leakage current.

8. The DRAM data retention time prediction method of claim 1, wherein, The preset lognormal distribution function comprises: a lognormal distribution cumulative function; a lognormal distribution probability density function; an exponential fitting function; a lognormal distribution cumulative density function and an exponential fitting function; or a lognormal distribution probability density function and an exponential fitting function.

9. The DRAM data retention time prediction method of claim 1, wherein, The collecting to-be-fitted data of the DRAM in a first preset time range of data retention time comprises: testing the DRAM to obtain a shmoo test result of the DRAM; obtaining a failure rate of the DRAM in a first preset time range of data retention time according to the shmoo test result, taking the failure rate as to-be-fitted data, and performing lognormal distribution fitting.

10. The DRAM data retention time prediction method of any of claims 1-9, wherein, The method further comprises: determining a process parameter for manufacturing the DRAM according to the key parameters.

11. A DRAM data retention time prediction apparatus, characterized by, The method comprises: a data collection module configured to collect to-be-fitted data of the DRAM in a first preset time range of data retention time; a data fitting module configured to fit the to-be-fitted data by using a preset lognormal distribution function to obtain key parameters, and obtain a target lognormal distribution function in a case where the key parameters meet a preset fitting condition; determining a data retention time of the DRAM according to the target lognormal distribution function; the key parameters include a mean, a standard deviation, a slope of the target lognormal distribution function in the first preset time range, and an intercept of the target lognormal distribution function on a time coordinate axis; the key parameters satisfying the preset fitting condition include: the standard deviation and the reciprocal of the slope are in a linear relationship, and a fitting goodness of the standard deviation and the reciprocal of the slope is greater than or equal to a preset precision threshold.

12. An electronic device, comprising: a processor and a memory, the memory being configured to store executable instructions of the processor; wherein the processor is configured to execute the executable instructions to perform the DRAM data retention time prediction method in any one of claims 1-10.

13. A computer readable storage medium having stored thereon a computer program, characterized in that, the computer program is executed by the processor to implement the DRAM data retention time prediction method in any one of claims 1-10.

Images