JP2011187713A - Overlay deviation error inspection method and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To optimize a correction factor, while maintaining the accuracy of sampling of an overlay deviation inspection. <P>SOLUTION: The overlay deviation inspection method comprises a process to determine a tolerance of a measured value, a process to determine a first sampling plan and calculation conditions for a first acceptance value, a process to obtain a first measured value, a process to obtain a first distribution function, an overlay deviation error factor, a first acceptance rate, and a first correction factor accuracy, a process to determine a second sampling plan based on the foregoing overlay deviation error factor and to determine calculation conditions for a second acceptance value, a process to obtain a second measured value, a process to obtain a second distribution function, a second acceptance rate, and a second correction factor accuracy, and a process to compare the first distribution function, the first acceptance rate, and the first correction factor accuracy with the second distribution function, the second acceptance rate, and the second correction factor accuracy respectively by using the tolerance, and then to make evaluation as to which is more suitable between the first sampling plan and the second sampling plan. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a misalignment error inspection method and program.

  The characteristics of a semiconductor integrated circuit are closely related to the dimensions of patterns formed in each manufacturing process and misalignment between layers. Therefore, in the manufacture of a semiconductor integrated circuit, an inspection process for monitoring whether or not the dimension and the misalignment between layers is controlled is provided. In this inspection process, since the number of patterns formed on the wafer is extremely large, it is difficult to monitor all the patterns, and therefore a sampling inspection is performed instead of a total inspection. In a general semiconductor integrated circuit manufacturing process, a set of a plurality of semiconductor chips formed on a single wafer or a plurality of wafers is manufactured as one product lot L. Accordingly, in the sampling inspection, a predetermined number of wafers are selected from the product lot L, and the position and number of chips to be inspected are selected from the selected wafers.

  However, because the process capability of the manufacturing process changes with time, when sampling inspection is performed using a certain sampling plan, the risk of considering a non-defective product as defective (hereinafter referred to as “α risk”) and the defective product are missed. Risk (hereinafter referred to as “β risk”) may fluctuate.

On the other hand, for example, a method of changing the sampling plan by changing the sample size or sampling location is conceivable. Common sampling inspection, hypergeometric distribution inspection characteristic curve based on (O perating C haracteristic C urves: hereinafter, referred to as "OC curve") count is calculated theoretically create and α risk and β risk sampling In addition to the inspection (for example, Non-Patent Document 1), the measurement value sampling inspection (for example, non-patent) for calculating the α risk and the β risk by theoretically creating an OC curve based on the non-central t distribution assumed to be a normal population Document 2) is known. In a general inspection process, the sampling inspection disclosed in Non-Patent Documents 1 and 2 is performed for a plurality of sampling plans.

  However, since the sampling inspections of Non-Patent Documents 1 and 2 are premised on random sampling, an OC curve cannot be created if this premise is not satisfied. In addition, since the measurement sampling in Non-Patent Document 2 is based on the premise that the population follows a normal distribution, an OC curve of the non-normal distribution of the population cannot be created. In other words, in the conventional semiconductor integrated circuit manufacturing process, if the assumption of random sampling or the assumption of the normal population is not established, the sampling inspection is not accurately evaluated, so the sampling inspection determined by the operator based on experience The inspection process is performed using As a result, α risk and β risk exceeding the allowable range may occur.

  In view of such circumstances, there is provided a method for accurately evaluating a sampling inspection even when a random sampling premise or a normal population premise is not satisfied (for example, Patent Document 1).

JP 2009-1769009 A

H. F. Dodge and H.C. G. Roming: “Single Sampling and Double Sampling Inspection Tables”, The BeL System Technical Journal, pp. 1-61 (1941) Military Standard: “Sampling Procedures and Tables for Inspection by Variables for Percent Detect: MIL-STD-414” (1957)

  The present invention provides a misalignment error inspection method and program capable of optimizing misalignment correction coefficients while maintaining sampling accuracy of misalignment inspection.

According to one aspect of the invention,
Determining an allowable range of measurement values in an inspection of alignment accuracy between the first pattern formed on the substrate and the second pattern transferred to the upper layer of the first pattern;
Determining a first sampling plan for selecting a first target for sampling inspection from a product lot and a condition for calculating a first pass judgment value;
Measuring misalignment in a product lot based on the first sampling plan and obtaining a first measurement value;
Obtaining a first distribution function, a misalignment error factor, a first pass rate and a first correction coefficient accuracy based on the calculation conditions of the first measurement value and the first pass determination value;
Determining a second sampling plan for selecting a second target for sampling inspection from the product lot based on the misalignment error factor, and determining a calculation condition of a second acceptance determination value;
Measuring misalignment in the product lot based on the second sampling plan and obtaining a second measurement value;
Obtaining a second distribution function, a second pass rate and a second correction coefficient accuracy based on the calculation conditions of the second measurement value and the second pass determination value;
Using the allowable range, comparing the first distribution function, the first pass rate and the first correction coefficient accuracy, and the second distribution function, the second pass rate and the second correction coefficient accuracy. To evaluate which of the first sampling plan and the second sampling plan is appropriate;
A process of sampling misalignment errors with a sampling plan that has been evaluated as appropriate;
A misalignment error inspection method is provided.

  According to the present invention, it is possible to provide a misalignment error inspection method and program capable of optimizing misalignment correction coefficients while maintaining the accuracy of sampling for misalignment inspection.

The figure which shows an example of the computer for performing the evaluation method of the misalignment error sampling inspection by one Embodiment of this invention. The flowchart which shows the schematic procedure of the evaluation method of the misalignment error sampling inspection by one Embodiment of this invention. The figure which shows an example of a 1st sampling plan. The table | surface which shows an example of 1st measurement data. The table | surface which shows an example of 1st calculation data. Explanatory drawing of a 1st distribution function and a 1st acceptance rate. The figure which shows the type | formula which calculates | requires the 1st correction coefficient calculation precision. The graph which plotted F value obtained by the analysis of variance from the 1st measured value. An example of the 2nd sampling plan created based on misalignment error factor analysis is shown. The figure which shows the example which overlaid the 1st and 2nd test | inspection characteristic curve. The figure which shows the example which displayed the calculation accuracy of the 1st and 2nd misalignment correction coefficient side by side.

  Hereinafter, some embodiments of the present invention will be described in detail with reference to the drawings. In the drawings, the same parts are denoted by the same reference numerals, and redundant description will be given only when necessary.

  FIG. 1 shows a computer 10 for executing an evaluation method for misalignment error sampling inspection according to an embodiment of the present invention. An external hard disk device 12 is connected to the computer 10, and this external hard disk device 12 has a plurality of recording areas and describes each procedure of evaluation of misalignment error sampling inspection described in detail below. In addition to the recipe file, various data in the middle of the inspection described later are stored. The computer 10 reads the recipe file from the hard disk device 12, evaluates the misalignment error sampling inspection, and outputs the evaluation result. The recording medium is not limited to a fixed recording medium such as the hard disk device 12 or a memory, and may be a portable medium such as a magnetic disk or an optical disk. The computer 10 is also connected to a misalignment error sampling inspection device 20. The misalignment error sampling inspection device 20 includes a misalignment measuring device 22 and an inspection unit 24. A first pattern P1 is formed on the wafer W to be inspected in the product lot L to be inspected, and a second pattern P2 is further formed on the first pattern P1. The misalignment measuring device 22 receives the wafer W carried from the product lot L, measures the misalignment of the first and second patterns at the inspection target location, and supplies the measurement result to the computer 10. The inspection unit 24 receives the output of the evaluation result of the misalignment error sampling inspection from the computer 10 and performs the misalignment error sampling inspection with an appropriate sampling plan.

  A schematic procedure of the evaluation method of the misalignment error sampling inspection according to the present embodiment will be described with reference to the flowchart of FIG.

  First, an allowable range of measured values is determined (step S201). For example, when the pass rate of the first sampling plan (hereinafter referred to as “first pass rate”) is determined to be 90% or more in α risk, the pass rate of the second sampling plan (hereinafter referred to as “second pass rate”). ")" Is determined to be 85% or more. Similarly for β risk, if the first pass rate is determined to be 10% or less, the second pass rate is determined to be 15% or less.

  Next, calculation conditions for the first sampling plan to be inspected and the pass determination value of the first sampling plan (hereinafter referred to as “first pass determination value”) are determined (step S202).

  Next, based on the determined first sampling plan, a measurement value of the product lot L (hereinafter referred to as “first measurement value”) is acquired (step S203).

  Next, based on the first pass determination value calculation condition determined in step S202, the first pass determination value is calculated from the first measurement value of the product lot L acquired in step S203, and the lot L number, sample Data including the wafer and the first pass determination value (hereinafter referred to as “first calculation data”) is written to the hard disk device 12 (S204).

  Subsequently, the first distribution function, the misalignment error factor, and the first pass rate are calculated based on the obtained first calculation data, and the calculation accuracy of the correction coefficient in the first sampling (hereinafter referred to as “first correction”). (Referred to as “coefficient calculation accuracy”) (step S205).

  Next, the second sampling plan to be inspected is determined based on the misalignment error factor, and further, the calculation condition of the pass determination value (hereinafter referred to as “second pass determination value”) of the second sampling plan is determined (step) S206).

  Next, based on the determined second sampling plan, the measurement value of the product lot L (hereinafter referred to as “second measurement value”) is acquired using the misalignment measuring device 22 (see FIG. 1) (step S207). .

  Next, based on the second pass determination value calculation condition determined in step S206, the second pass determination value is calculated from the second measurement value of the product lot L acquired in step S207, and the lot L number, the sample wafer are calculated. Then, data including the first pass determination value (hereinafter referred to as “second calculation data”) is written in the hard disk device 12 (step S208).

Subsequently, the second pass rate is calculated based on the obtained second calculation data, and the calculation accuracy of the correction coefficient in the second sampling (hereinafter referred to as “second correction coefficient calculation accuracy”) is obtained (step S209). )
Finally, by comparing the first pass rate and the second pass rate, the first correction coefficient calculation accuracy and the second correction coefficient calculation accuracy, it is evaluated which of the first sampling plan and the second sampling plan is appropriate. (Step S210).

  Subsequently, each procedure after step S202 described above will be specifically described.

  First, in step S202, the first sampling plan for misalignment inspection includes, for example, the number N of wafers W (hereinafter referred to as “sample wafers”) W to be extracted from the product lot L and the position of the sample wafer W in the lot L (hereinafter referred to as “sample wafer”). This is called “placement in a lot”. The arrangement in the lot is the number of the sample wafer W and the position of the chip to be inspected (hereinafter referred to as “sample chip”) on the wafer W and the inspection position in the chip. More specifically, for example, as shown in FIG. 3, wafer numbers 1 to 3 are determined as sample wafers, and five chips included in the sample wafer are selected as sample chips. Are selected as inspection positions. FIG. 3 shows a case where 10% or less is determined as the first pass rate. In addition, the calculation condition of the first pass determination value is, for example, when the measurement value of the sample wafer is compared with the standard, and when the minimum value of the measurement value is out of the standard range, The minimum measured value is a condition for calculating the first pass judgment value. In addition, when the measured value of the sample wafer is compared with the standard and the third largest value out of the measured value is out of the standard range, the third largest value is the first. This is a condition for calculating a pass judgment value.

  Next, in step S203, the first measurement value is acquired by using the misalignment measuring apparatus 22 (see FIG. 1), for example, in the example of FIG. , The value of misalignment error is obtained by measuring 20 points (5 chips × 4 locations). At this time, as shown in FIG. 4, the computer 10 has data (hereinafter referred to as “first measurement data”) including the lot L number, wafer number, sample chip position, and pattern type corresponding to the first measurement value. ) Is written to the hard disk device 12. An example of the first measurement data is shown in FIG. The first measurement data in FIG. 4 includes the lot L number, wafer number, sample chip position, and pattern type corresponding to the first measurement value. In general, the larger the number of product lots L determined as sample wafers, the more statistically preferable first measurement value is obtained. The smaller the number of product lots L determined as sample wafers, the smaller the number of product lots L is. A first measurement value with small variation is obtained.

  Subsequently, an example of the first calculation data obtained in step S204 is shown in the table of FIG. The table of FIG. 6 shows, for each product lot L, the first measurement value for each sample wafer and the first calculation data including the maximum value and the minimum value.

Next, the first distribution function and the first pass rate calculated in step S205 will be described with reference to FIG. First, the histogram of the first calculation data calculated in step S204 is calculated, the histogram is smoothed using the kernel density estimation method, and the probability density function of the maximum value and the minimum value is obtained from the smoothed histogram. Output as a distribution function. Subsequently, the probability density function is numerically integrated (see the hatched portion in FIG. 6) to calculate a probability distribution function, and an upper probability and a lower probability at arbitrary values are calculated as the first pass rate. Thus, the first inspection characteristic curve is calculated. On the other hand, the calculation accuracy of the correction coefficient in the first sampling (hereinafter referred to as “first correction coefficient calculation accuracy”) is calculated based on the first calculation data using the equations (1) and (2) shown in FIG. Ask. In FIG. 7, T _X and T _Y represent the shift amounts of the entire pattern in the X direction and the Y direction, respectively. Further, _{M X} and _{M Y} represents a magnification of the pattern in the X and Y directions. Further, R and S represent the rotation angle of the pattern and the degree of skew, respectively.

  Next, a method for determining the second sampling plan will be described. From the first measured value, analysis of variance (ANOVA) is used to analyze which of the wafer factor, shot factor, and site factor is greater as a factor of misalignment error. FIG. 8 is a graph plotting the F value of such an analysis of variance. FIG. 8 shows that the shot factor is the largest and the wafer factor is small. Therefore, in the present embodiment, when creating the second sampling plan, it is selected to reduce the number of wafers compared to the first sampling plan.

FIG. 9 shows an example of a second sampling plan created based on such an analysis of variance. In the example shown in FIG. 9, two wafers W are extracted from the product lot L, and for each wafer W, predetermined four locations (that is, 8) of two chips out of 20 points of the first sampling plan determined in step S202. Twelve points excluding (point) are selected as inspection positions. Here, the calculation condition of the second pass determination value is, for example, Expression 3.

はサンプルの平均値であり、ステップＳはサンプルの標準偏差であり、ｋは合格判定係数（例えば、ｋ＝２．０）である。複号は、上限に対して＋、下限に対して−である。なお、サンプルの平均値に代えて、サンプルの中央値が用いられても良い。 In Equation 3,

Is an average value of the sample, step S is the standard deviation of the sample, and k is a pass determination coefficient (for example, k = 2.0). The compound number is + for the upper limit and-for the lower limit. Note that the median value of the samples may be used instead of the average value of the samples.

  If an algorithm for predicting the inspection time is incorporated in the recipe file, a plurality of second sampling candidates are set, the inspection time is calculated for each candidate, and the candidate with the shortest inspection time is determined as the second sampling plan. You can also

  Thereafter, as in the first sampling plan described above, the second measurement value of the product lot L is acquired for the second sampling plan, and the lot L number, wafer number, and sample chip position corresponding to the second measurement value are acquired. The data including the pattern type (hereinafter referred to as “second measurement data”) is written in the hard disk device 12 (step S207). Subsequently, based on the calculation condition of the second pass determination value, the second pass determination value is calculated from the second measurement value, and data including the lot L number, the sample wafer, and the second pass determination value (hereinafter referred to as “second”). Is written in the hard disk device 12 (step S208).

  Further, based on the second calculation data, the second distribution function and the second pass rate are calculated in the same procedure as in step S204 described above, a second inspection characteristic curve is obtained, and the second calculation data is calculated based on FIG. The calculation accuracy of the correction coefficient in the second sampling (hereinafter referred to as “second correction coefficient calculation accuracy”) is obtained using the calculation formula (step S209).

  In the next step S210, the calculation results of steps S205 and S209 described above are combined. For example, as shown in FIG. 10, the first and second inspection characteristic curves are overwritten in a common coordinate system. The superimposed calculation results are compared with each other using the allowable range determined in step S201. Further, as shown in FIG. 11, the calculation accuracy of the first and second misalignment correction coefficients is displayed side by side to be compared with each other. Based on these comparison procedures, it is evaluated which one of the first and second sampling plans is suitable for the sampling inspection. For example, in the overlay result shown in FIG. 10, the α risk and β risk of the second test characteristic curve are significantly larger than the α risk and β risk of the first test characteristic curve, respectively. Indicates that it should not be changed to the second sampling plan. Further, as shown in FIG. 11, since the calculation accuracy of the first and second misalignment correction coefficients is significantly different, it is found that the first sampling plan should not be changed to the second sampling plan.

  According to the present embodiment, even if the assumption of random sampling or the assumption of the normal population is not established, the sampling inspection can be accurately evaluated, and consequently, the inspection process is performed using an appropriate sampling inspection method. Thus, the misalignment correction coefficient can be optimized while maintaining the sampling accuracy of the misalignment inspection.

  Then, if the inspection result by the appropriate sampling inspection method is reflected in the subsequent manufacturing lot L in this way, the semiconductor device can be manufactured with high yield and throughput.

10: Computer 12: Hard disk device 20: Misalignment error sampling inspection device 22: Misalignment measuring device 24: Inspection section L: Product lot L
P1: First pattern P2: Second pattern W: Wafer

Claims (4)

Determining an allowable range of measurement values in an inspection of alignment accuracy between the first pattern formed on the substrate and the second pattern transferred to the upper layer of the first pattern;
Determining a first sampling plan for selecting a first target for sampling inspection from a product lot and a condition for calculating a first pass judgment value;
Measuring misalignment in a product lot based on the first sampling plan and obtaining a first measurement value;
Obtaining a first distribution function, a misalignment error factor, a first pass rate and a first correction coefficient accuracy based on the calculation conditions of the first measurement value and the first pass determination value;
Determining a second sampling plan for selecting a second target for sampling inspection from the product lot based on the misalignment error factor, and determining a calculation condition of a second acceptance determination value;
Measuring misalignment in the product lot based on the second sampling plan and obtaining a second measurement value;
Obtaining a second distribution function, a second pass rate and a second correction coefficient accuracy based on the calculation conditions of the second measurement value and the second pass determination value;
Using the allowable range, comparing the first distribution function, the first pass rate and the first correction coefficient accuracy, and the second distribution function, the second pass rate and the second correction coefficient accuracy. To evaluate which of the first sampling plan and the second sampling plan is appropriate;
A process of sampling misalignment errors with a sampling plan that has been evaluated as appropriate;
A misalignment error inspection method comprising:   The method of claim 1, wherein the second sampling plan is determined by deleting a part of the measurement points of the first sampling plan.   The second sampling plan calculates a plurality of second sampling plan candidates from which different portions are deleted from the measurement points of the first sampling plan, and the inspection time is the shortest among the plurality of second sampling plan candidates. 3. The misalignment error inspection method according to claim 1, wherein the error is determined by selecting a candidate. A procedure for determining an allowable range of measurement values in an inspection of alignment accuracy between the first pattern formed on the substrate and the second pattern transferred to the upper layer of the first pattern;
A procedure for determining a first sampling plan for selecting a first target for sampling inspection from a product lot and a condition for calculating a first acceptance judgment value;
Measuring a misalignment in a product lot based on the first sampling plan and obtaining a first measurement value;
A procedure for obtaining a first distribution function, a misalignment error factor, a first pass rate, and a first correction coefficient accuracy based on the calculation conditions of the first measurement value and the first pass determination value;
Determining a second sampling plan for selecting a second target for sampling inspection from the product lot based on the misalignment error factor, and determining a calculation condition for a second acceptance determination value;
Measuring a misalignment in the product lot based on the second sampling plan and obtaining a second measurement value;
A procedure for obtaining a second distribution function, a second pass rate and a second correction coefficient accuracy based on the calculation conditions of the second measurement value and the second pass determination value;
Using the allowable range, comparing the first distribution function, the first pass rate and the first correction coefficient accuracy, and the second distribution function, the second pass rate and the second correction coefficient accuracy. To evaluate which of the first sampling plan and the second sampling plan is appropriate;
A procedure to perform misalignment error sampling inspection with a sampling plan evaluated as appropriate,
A program for causing a computer to perform an evaluation of a misalignment error sampling inspection.

Images