CN105702595A - Wafer yield judging method and wafer qualification test multivariable detection method - Google Patents

Abstract

The invention provides a wafer yield judging method and a wafer qualification test multivariable detection method. The yield rate judging method comprises the following steps: carrying out wafer qualification test on the wafer to be tested to generate a plurality of characteristic parameters; grouping the characteristic parameters according to the correlation list to form a plurality of groups of characteristic parameter groups; calculating an analysis value corresponding to one or a combination of the characteristic parameters in each group of characteristic parameter groups by multivariate analysis, and judging whether the analysis value is larger than preset specification information corresponding to one or a combination of the characteristic parameters, so as to take one or a combination of the characteristic parameters corresponding to the analysis value larger than the preset specification information as at least one out-of-specification characteristic parameter; and judging the yield of the wafer to be tested according to the out-of-specification characteristic parameters.

Description

Wafer Yield Judgment Method and Wafer Qualification Test Multivariate Detection Method

technical field

The present invention relates to a wafer inspection technique, and in particular relates to a wafer yield judgment method and a multiple variable (Multiple variable) wafer acceptance test (WAT) for wafer acceptance test. )Detection method.

Background technique

During semiconductor fabrication, wafer qualification testing (WAT) is commonly used to perform electrical testing on wafers. WAT will generate many electrical parameters and values. Users can use these electrical parameters and values to evaluate the yield of the wafer under test and discover possible problems in the semiconductor manufacturing process. If the abnormal WAT parameters can be found early, it can have a decisive impact on the decision-making of wafer manufacturing, the evaluation of whether the wafer needs to be scrapped, and whether it will be shipped to the customer.

As far as the current technology is concerned, the electrical parameters and values generated by WAT are checked manually by wafer researchers to find out problems from these electrical parameters. For example, the wafer qualification testing system can automatically generate a list of WAT parameters, and researchers check whether these WAT parameters exceed the preset range (Outofspecification, OOS for short). If the WAT parameters have exceeded the preset range, the researchers will check whether these WAT parameters beyond the preset range affect the yield of the wafer, and then evaluate the wafer according to the severity of the impact of the WAT parameters on the wafer yield Whether it needs to be scrapped.

However, the types of WAT parameters are very complicated and may reach more than 300. It is not easy to find problems in WAT parameters manually. On the other hand, different WAT parameters may be correlated with each other; when a certain parameter exceeds the preset range, other parameters related to this parameter may also exceed the preset range, but this correlation is difficult to pass easily. Human confirmation. Therefore, if the WAT parameters are detected manually, a lot of manpower and precious time will be spent. Therefore, it is necessary to find a wafer yield detection and testing technology that can quickly detect WAT parameters and can greatly save labor costs.

Contents of the invention

The invention provides a method for judging the yield rate of wafers and a multivariate detection method for qualified testing of wafers. This yield rate judgment method uses multivariate statistical process control (MSPC for short) and the assistance of hardware equipment to list WAT characteristic parameters beyond the preset range, so as to reduce the detection time of R&D personnel for WAT characteristic parameters, and save time. The test cost for judging the yield of the wafer under test.

The invention proposes a wafer yield judgment method, which includes the following steps. Wafer qualification testing is performed on the wafer under test to generate a number of characterization parameters. The characteristic parameters are grouped according to a correlation list to form a plurality of characteristic parameter groups. Use multivariate analysis to calculate the analysis value corresponding to one of these characteristic parameters or a combination of these characteristic parameters in each group of characteristic parameters, and determine whether the analysis value is greater than the preset value corresponding to one of these characteristic parameters or a combination thereof Specification information, using one or a combination of these characteristic parameters corresponding to an analysis value greater than the preset specification information as at least one out-of-specification characteristic parameter. And, judge the yield rate of the wafer to be tested according to the at least one out-of-spec characteristic parameter.

In an embodiment of the present invention, the above yield judgment method further includes the following steps. The characteristic parameters produced by a plurality of different wafers after performing the wafer qualification test are counted to analyze whether these characteristic parameters are correlated. And, the correlation list is generated according to a plurality of correlation characteristic parameters with correlation.

In an embodiment of the present invention, the above yield judgment method further includes the following steps. A plurality of related property parameters having dependencies among these property parameters are recorded to form this dependency list.

In an embodiment of the present invention, the above-mentioned multivariate analysis uses Hotelling's T-squared statistic (Hotelling's T-squared statistic) to sequentially calculate one or a combination of these characteristic parameters in each group of characteristic parameters.

In one embodiment of the present invention, calculating the analysis value corresponding to one or a combination of these characteristic parameters in each characteristic parameter group includes the following steps: sequentially setting these characteristic parameters in each characteristic parameter group N of them are used as variables to calculate whether each N-dimensional T-square statistical value in each characteristic parameter group is greater than the corresponding preset specification information, wherein N is a positive integer.

In one embodiment of the present invention, judging the yield rate of the wafer to be tested according to the at least one out-of-spec characteristic parameter includes the following steps: listing the at least one out-of-spec characteristic parameter as an out-of-spec parameter list, to It is for the user to judge whether the wafer to be tested is abnormal.

From another point of view, the present invention proposes a multivariate detection method for wafer qualification testing, which is applicable to a wafer qualification testing system. This multivariate judgment method includes the following steps: Multiple characteristic parameters generated after testing. Group these characteristic parameters according to the correlation list to form multiple groups of characteristic parameters; use multivariate analysis to calculate whether the analysis value corresponding to one of these characteristic parameters or their combination in each group of characteristic parameters is greater than The preset specification information corresponding to one or a combination of these characteristic parameters, so that one or a combination of these characteristic parameters corresponding to an analysis value greater than the preset specification information is regarded as at least one out-of-specification characteristic parameter and, checking whether the wafer to be tested is qualified according to the at least one out-of-spec characteristic parameter.

Based on the above, the wafer yield judgment method and the multivariate detection method of the wafer qualification test in the embodiment of the present invention group a plurality of WAT characteristic parameters with correlation, and use a multivariate statistical algorithm (for example, Hodlin T square statistics) to test each group of characteristic parameters, thereby screening out multiple WAT characteristic parameters beyond the preset range. In this way, the embodiment of the present invention can detect abnormal WAT characteristic parameters as early as possible through the multivariable manufacturing process control technology and the assistance of hardware equipment, reduce the detection time of R&D personnel for WAT characteristic parameters, and save the time spent on judging the wafer to be tested. Yield test cost.

In order to make the above-mentioned features and advantages of the present invention more comprehensible, the following specific embodiments are described in detail with reference to the accompanying drawings.

Description of drawings

FIG. 1 shows a flow chart of a method for judging the yield of a wafer according to an embodiment of the present invention;

Fig. 2 shows a detailed flowchart of generating a correlation list according to an embodiment of the present invention;

Fig. 3 is a two-dimensional sample schematic diagram of two characteristic parameters under multivariate analysis;

Fig. 4 to Fig. 6 show an embodiment of the present invention by adopting the experimental result analysis chart of the method for judging the yield of this wafer;

FIG. 7 shows a flow chart of a multivariate detection method for wafer qualification testing according to an embodiment of the present invention.

Explanation of reference signs:

S110～S140, S210～S240, S710～S740: steps;

310, 320, 330, 410, 510, 610: area;

340: preset range;

Y1～Y5, Y1_310, Y2_320: characteristic parameters;

360 : the point where the characteristic parameters Y1_310 and Y2_320 appear in the area 330 .

detailed description

FIG. 1 shows a flowchart of a wafer yield judgment method according to an embodiment of the present invention. Please refer to FIG. 1 , the so-called wafer is the carrier and material used to manufacture semiconductor integrated circuits in the semiconductor manufacturing process. Each wafer on the wafer is a product of the semiconductor manufacturing process. In order to save labor costs, the method for judging the yield rate of wafers and the multivariate detection method for wafer qualification testing described in the embodiments of the present invention can be presented in the form of software (for example, application programs), and equipped with better hardware The electronic equipment of the standard can execute this software, thereby realizing the embodiment of the present invention. This software can also be recorded on non-volatile readable media, such as optical discs, removable hard disks and other devices. Those who apply this embodiment can also implement the wafer yield judgment method and the multivariate detection method of wafer qualification test described in the embodiment of the present invention by means of firmware or hardware, and are not limited to the content of this disclosure.

Here, the method for judging the yield of a wafer is described using the step flow in FIG. 1 . In step S110, a wafer qualification test (WAT) is performed on the wafer to be tested to generate a plurality of characteristic parameters. Wafer qualification testing can use one or more procedures to obtain relevant parameters of various electrical and physical properties as characteristic parameters. These programs are, for example, wafer sort (Wafersort), wafer final test (Wafer final test), electronic circuit chip sort (Electronic diesort), and circuit probe (Circuit Probe). In this embodiment, the wafer probe is mainly used to perform the wafer qualification test on the wafer to be tested. Those who apply this embodiment can also use other types of procedures to perform wafer qualification testing.

After obtaining the characteristic parameters, because many WAT characteristic parameters are correlated, if these correlated WAT characteristic parameters are divided into the same group, and within these groups, the multivariate production process control ( It is more efficient to use MSPC) to integrate WAT characteristic parameters and determine wafer yield. The "correlation between WAT characteristic parameters" referred to here may mean that these WAT characteristic parameters are positively correlated, negatively correlated or any correlation described in statistics. Therefore, in step S120 , the embodiment of the present invention groups these characteristic parameters according to a correlation list, so as to form a plurality of characteristic parameter groups. The correlation list can be pre-set and generated through user experience, and can also be statistically obtained from a large amount of data of WAT characteristic parameters.

Fig. 2 shows a detailed flow chart of generating a correlation list according to an embodiment of the present invention. Referring to FIG. 2 , in step S210 , the user can record the relevant characteristic parameters according to his experience. Then, in step S240, these recorded characteristic parameters with correlations are integrated into a correlation list. On the other hand, in step S220, the characteristic parameters generated by a large number of wafers after the above-mentioned wafer qualification test can also be counted, so as to analyze whether these characteristic parameters are correlated. When a plurality of relevant characteristic parameters that are suspected to be relevant are obtained through statistical methods, it will enter step S230, and these relevant characteristic parameters that are suspected to be relevant can be formed into a list and provided to the user for viewing, so that the user can judge whether these relevant characteristic parameters are suspected to be relevant. Whether these correlation characteristic parameters of correlation are really correlated. When the user judges that the relevant characteristic parameters are indeed relevant according to his experience, the user proceeds from step S230 to step S240 to integrate the relevant characteristic parameters to form a correlation list.

Please continue to refer to FIG. 1. After these characteristic parameters are divided into multiple groups of characteristic parameters according to the correlation list, in step S130, multivariate analysis is used to calculate one of the characteristic parameters in each group of characteristic parameters. One or the analysis value corresponding to the combination thereof, and then compare the analysis value with the preset specification information, so as to determine whether one or the combination of these characteristic parameters in each characteristic parameter group exceeds the preset specification.

Here, FIG. 3 is used to illustrate why it is necessary to use multivariate analysis to perform wafer yield judgment on these characteristic parameters in each characteristic parameter group. Fig. 3 is a schematic diagram of a two-dimensional sample of two characteristic parameters under multivariate analysis. To simplify the description, FIG. 3 only has two characteristic parameters Y1 and Y2 of a plurality of wafer samples. The characteristic parameter Y1 is presented in area 310 and the characteristic parameter Y2 is presented in area 320 . The area 310 and the area 320 respectively indicate the maximum value, the minimum value and the middle value of the characteristic parameter Y1 and the characteristic parameter Y2 within the preset range. Therefore, if the characteristic parameter Y1 and the characteristic parameter Y2 are analyzed by means of univariate analysis, then these wafer samples should all be consistent. However, if the characteristic parameter Y1 and the characteristic parameter Y2 are presented in the area 330 in the form of negative correlation, it will be found that the point 360 where the characteristic parameters Y1_310 and Y2_320 of a certain wafer sample appear in the area 330 is not located in the characteristic The parameter Y1 and the characteristic parameter Y2 are within the preset range 340 as two-dimensional variables, so it can be known that the wafer sample may have a problem and needs to be scrapped. Therefore, the multivariate analysis can improve the judgment accuracy of the yield rate of the wafer, and the wafer to be tested with problems can be easily obtained through the yield judgment method according to the embodiment of the present invention.

Returning to FIG. 1 , in this embodiment, the "multivariate analysis" referred to in the embodiment of the present invention can be realized by Hotelling's T-squared statistic in statistical algorithms. However, the multivariate analysis can be realized through various implementation methods and statistical algorithms, and the embodiments of the present invention are not limited to Hodlin's T-square statistics. Step S130 of this embodiment can be subdivided into a plurality of detailed steps S132-S138, and here each characteristic parameter in each characteristic parameter group is regarded as a different variable. Hodlin's T square statistics can also be called MYT algorithm, which uses selected variables to calculate T square (T ² ) values for other variables under given conditions. Then, the T ² value is compared with the preset specification value (marked as T _spec ) under the same given conditions, so as to judge whether the selected variables are out of preset range (OOS). If the ^T2 values corresponding to these selected variables do not exceed the preset range, then use the recursive analysis method to sequentially select other selected variables, and repeat the above process until the T value corresponding to the selected variable ² until the value exceeds the preset range. Thereby, the yield rate of the wafer can be judged by one or a combination of the characteristic parameters corresponding to the T ² value exceeding the preset range.

Each detailed flow of steps S132 to S138 will be described in detail here. In step S132, N of these characteristic parameters in each characteristic parameter group are sequentially set as selected variables. N is a positive integer, and N can start counting from 1, and the maximum value of N is the number of multiple characteristic parameters in the set of characteristic parameter groups. In this embodiment, each characteristic parameter is called a different dimension, and when one of the characteristic parameters is given a condition, the T2 value calculated at this time is called a ¹ -dimensional T2 statistical value; when ^two of them When a characteristic parameter is given a condition, the T2 value calculated at this time is called a ^{2 -} dimensional T2 statistical value, and so on. In step S133, each N-dimensional T ² statistical value in each characteristic parameter group is calculated by using the selected variables under given conditions. Those who apply this embodiment can know the calculation method of the T ² statistical value through the Hodlin's T-square statistical algorithm or other statistical algorithms in accordance with the embodiments of the present invention. In step S134, it is determined whether each N-dimensional T ² statistical value is greater than the corresponding preset rule information (T _spec ). If the N-dimension T ² statistical values at this time are not greater than the corresponding preset rule information, then enter step S135 from step S134 to judge whether N is equal to the maximum value of N (the maximum value of N is also the characteristic parameter group of this group) number of multiple feature parameters). If N is not equal to the maximum value, the value of N is increased by one from step S136, and the corresponding processes of steps S132-S136 are repeated. Relatively, if N is equal to its maximum value, it means that the ^T2 values corresponding to one of the characteristic parameters or their combinations in the set of characteristic parameter groups do not have OOS, and thus enter step S137 from step S135, It is judged that the yield of the wafer to be tested is good.

If one of the N-dimensional T ² statistical values is greater than the corresponding preset rule information (T _spec ), then proceed from step S134 to step S138, and will be greater than the analysis value (T ² statistical value) of the preset specification information (T _spec ) One or a combination of the corresponding characteristic parameters is regarded as at least one out-of-spec characteristic parameter. In other words, since the analysis value corresponding to one of these characteristic parameters or a combination thereof is greater than the preset rule information (T _spec ), it means that one of these characteristic parameters or a combination thereof has exceeded the preset rule (OOS) . Thereby, in the embodiment of the present invention, one or a combination of these characteristic parameters exceeding the predetermined rule (OOS) is set as the out-of-spec characteristic parameter. In some embodiments, Hodlin's T ^- square statistical algorithm can also sum up all the analysis values (T2 statistical values) under the selected variables, and compare the total value with the preset information, so as to judge Whether the analysis value under this selected variable still has other analysis values that are too significant and may be problematic. If the total value of the analysis values under the selected variables is considered to be still problematic (or whether the total value is significant), the embodiment of the present invention can continue to execute steps S132-S138, so as to obtain more out-of-spec Feature parameters. The embodiment of the present invention uses the steps S132-S138 to briefly describe the spirit of the present invention. Those who apply this embodiment can adjust the sequence of steps S132-S138 according to their needs, and are not limited to the above disclosure.

In step S140, the yield rate of the wafer to be tested is judged according to the out-of-spec characteristic parameters found above. In this embodiment, step S140 can use various methods to determine the yield of the wafer to be tested according to these out-of-spec characteristics. One method may be to calculate these out-of-specification characteristic parameters through the wafer qualification test and judgment system, and judge whether the wafer to be tested is abnormal according to the calculation and judgment results of the wafer qualification test and judgment system. Another approach is to list these specifications and characteristic parameters as a list (for example, beyond the rule parameter list), so that users can judge whether these WAT characteristic parameters are abnormal based on their experience, so as to judge the yield of the wafer to be tested .

FIG. 4 to FIG. 6 show analysis diagrams of experimental results using the wafer yield judgment method according to an embodiment of the present invention. 4 to 6 respectively present the characteristic parameters Y3, Y4 and Y5 of the wafer. Marks SpecH and SpecL are respectively the maximum value (such as 4.1, -10 and 1) and minimum value (such as 2.3, -40 and 0.9) of this specific parameter in the preset range corresponding to the characteristic parameters Y3, Y4 and Y5 respectively . Thus, it can be seen from the areas 410, 510, and 610 that if the yield rate of the wafer is judged by the conventional univariate analysis, since the characteristic parameters Y3, Y4, and Y5 in the areas 410, 510, and 610 are all within the preset range Instead, some wafers are considered to have good yields. However, if the yield rate of the wafer is judged by multivariate analysis, these values in the regions 410 , 510 , and 610 will actually exceed the preset range due to other variables.

From another perspective, the embodiment of the present invention proposes a multivariate detection method for wafer qualification testing, which is applicable to a wafer qualification testing system. The wafer qualification testing system may include a testing platform for measuring various WAT electrical parameters of the wafer and an operation module for implementing the multivariate testing method. The computing module can realize the embodiment of the present invention through the step flow in FIG. 7 . FIG. 7 shows a flow chart of a multivariate detection method for wafer qualification testing according to an embodiment of the present invention. In step S710, the computing module obtains a plurality of characteristic parameters generated after the wafer to be tested passes the test platform to perform the wafer qualification test. In step S720, the operation module groups these characteristic parameters according to the correlation list to form a plurality of characteristic parameter groups. In step S730, the operation module uses multivariate analysis to calculate the analysis value corresponding to one or a combination of these characteristic parameters in each group of characteristic parameters, and determines whether the analysis value is greater than one of these characteristic parameters or the preset specification information corresponding to the preset specification information, so that one or a combination of the characteristic parameters corresponding to the analysis value greater than the preset specification information is regarded as at least one out-of-specification characteristic parameter. In step S740, the calculation module detects whether the wafer to be tested is qualified according to the at least one out-of-spec characteristic parameter. For the detailed flow of steps S710-S740, please refer to the above-mentioned embodiments.

In summary, the method for judging the yield rate of wafers and the multivariate detection method for wafer qualification testing in the embodiments of the present invention group multiple WAT characteristic parameters with correlation, and use multivariate statistical algorithms (for example, Hodlin's T square statistics) to test the characteristic parameters of each group, so as to screen out multiple WAT characteristic parameters beyond the preset range. In this way, the embodiment of the present invention can detect abnormal WAT characteristic parameters as early as possible through the multivariable manufacturing process control technology and the assistance of hardware equipment, reduce the detection time of R&D personnel for WAT characteristic parameters, and save the time spent on judging the wafer to be tested. Yield test cost.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than limiting them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: It is still possible to modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the technical solutions of the various embodiments of the present invention. scope.

Claims (12)

1. A method for judging the yield rate of a wafer, characterized in that, comprising: Perform wafer qualification testing on the wafer to be tested to generate multiple characteristic parameters; Grouping these characteristic parameters according to the correlation list to form multiple sets of characteristic parameter groups; Multivariate analysis is used to calculate the analysis value corresponding to one of the characteristic parameters or a combination of these characteristic parameters in each group of characteristic parameters, and to determine whether the analysis value is greater than the corresponding one of the characteristic parameters or a combination thereof Preset specification information, so that one or a combination of the characteristic parameters corresponding to the analysis value greater than the preset specification information is used as at least one out-of-specification characteristic parameter; and The yield rate of the wafer to be tested is judged according to the at least one out-of-spec characteristic parameter. 2. the yield judgment method of wafer according to claim 1, is characterized in that, also comprises: making statistics on the characteristic parameters produced by a plurality of different wafers after the wafer qualification test, to analyze whether the characteristic parameters are relevant; and The correlation list is generated according to a plurality of correlation characteristic parameters having the correlation. 3. the yield judgment method of wafer according to claim 1, is characterized in that, also comprises: A plurality of correlated characteristic parameters among the characteristic parameters are recorded to form the correlation list. 4. The method for judging the yield of a wafer according to claim 1, wherein the multivariate analysis uses Holderlin's T square statistics to sequentially calculate one of the characteristic parameters in each group of characteristic parameters or its combination. 5. The method for judging the yield of a wafer according to claim 4, wherein calculating the analysis value corresponding to one of the characteristic parameters in each group of characteristic parameters or a combination thereof comprises the following steps: Sequentially setting N of these characteristic parameters in each characteristic parameter group as selected variables, to calculate each N-dimensional T-square statistical value in each characteristic parameter group; It is judged whether each N-dimensional T-square statistical value is greater than the corresponding preset specification information, wherein N is a positive integer. 6. The method for judging the yield of the wafer according to claim 1, wherein judging the yield of the wafer to be tested according to the at least one out-of-spec characteristic parameter comprises the following steps: The at least one out-of-standard characteristic parameter bar is listed as an out-of-rule parameter list for users to judge whether the wafer to be tested is abnormal. 7. A multivariate detection method for wafer qualification testing, suitable for wafer qualification testing systems, characterized in that the multivariate judgment method comprises: Obtain multiple characteristic parameters of the wafer to be tested after the wafer qualification test; Grouping these characteristic parameters according to the correlation list to form multiple sets of characteristic parameter groups; Multivariate analysis is used to calculate the analysis value corresponding to one of the characteristic parameters or a combination of these characteristic parameters in each group of characteristic parameters, and to determine whether the analysis value is greater than the corresponding one of the characteristic parameters or a combination thereof Preset specification information, so that one or a combination of the characteristic parameters corresponding to the analysis value greater than the preset specification information is used as at least one out-of-specification characteristic parameter; and Whether the wafer to be tested is qualified or not is detected according to the at least one out-of-spec characteristic parameter. 8. multivariate detection method according to claim 7, is characterized in that, also comprises: making statistics on the characteristic parameters produced by a plurality of different wafers after the wafer qualification test, to analyze whether the characteristic parameters are relevant; and The correlation list is generated according to a plurality of correlation characteristic parameters having the correlation. 9. multivariate detection method according to claim 7, is characterized in that, also comprises: A plurality of correlated characteristic parameters among the characteristic parameters are recorded to form the correlation list. 10. The multivariate detection method according to claim 7, wherein the multivariate analysis uses a Hodlin's T square statistic to sequentially calculate one of the characteristic parameters in each group of characteristic parameter groups or a combination thereof . 11. The multivariate detection method according to claim 10, wherein calculating the analysis value corresponding to one or a combination of the characteristic parameters in each group of characteristic parameters comprises the following steps: Sequentially setting N of these characteristic parameters in each characteristic parameter group as variables to calculate whether each N-dimensional T-square statistical value in each characteristic parameter group is greater than the corresponding preset specification information, Where N is a positive integer. 12. The multivariate detection method according to claim 7, wherein judging the yield rate of the wafer to be tested according to the at least one out-of-spec characteristic parameter comprises the following steps: The at least one out-of-standard characteristic parameter bar is listed as an out-of-rule parameter list for users to judge whether the wafer to be tested is qualified or not based on their user experience.