JP5078151B2 - Defect detection level adjustment method and defect inspection system - Google Patents

Description

  The present invention relates to a defect detection level adjustment method and a defect inspection system for an inspection object, and in particular, a defect detection level adjustment capable of optimizing a defect detection level of an inspection object such as a photomask or a semiconductor wafer. The present invention relates to a method and a defect inspection system.

  In order to guarantee the quality of products such as photomasks and semiconductor wafers, it is necessary to detect defects present in these products, determine whether products are acceptable, and correct defects as necessary. The accuracy of this defect detection is determined by a sensitivity parameter set for one or more defect detection algorithms for detecting defects.

Here, the sensitivity parameter is a threshold value for the defect inspection system to determine that a portion having a non-inspection object is a defect. For example, when the value detected as the difference amount in the comparative inspection method is larger than a preset sensitivity parameter, the location is detected as a defect.

  If this sensitivity parameter is set to an excessively high sensitivity in consideration of the design safety factor, defects that are not originally required to be detected as defects (hereinafter referred to as “newsence defects”) are detected as actual defects. As a result, a quality mask or the like is manufactured, and as a result, the yield is lowered.

  Therefore, it is important to set the optimum sensitivity parameter before performing defect detection. The optimum sensitivity parameter setting is complicated by pattern type, technology node, RET (Resolution Enhancement Technologies) processing, etc. Because it varies depending on the degree, it cannot be set uniformly.

Further, when performed twice or more defect detection for the same inspected object at a predetermined sensitivity parameters set, ideally, although the rate of detection of the defect should show a constant value, actual The size of the defect to be inspected is as small as nm order, so the vibration of the defect detection device, optical system deterioration, light source fluctuation, stage running accuracy (yawing and pitching), stage stabilization accuracy The detection rate may fluctuate every time a defect is detected due to various factors such as atmospheric pressure fluctuation and temperature fluctuation.

  Therefore, in order to optimize the sensitivity parameter, it was designed according to the properties of the inspection method such as the die comparison method, the transmitted light and reflected light illumination method, the inspection pattern of the object to be inspected, and the material. After selecting a defect detection algorithm, a defect inspection device (actual machine) actually creates multiple inspected objects such as masks based on specific data consisting of defect size data, coordinate data, etc. It was necessary to set the optimum sensitivity parameters through trial and error while detecting the program defects of the above and confirming the defect detection status by review.

  FIG. 7 is a processing flow showing a procedure for optimizing the general sensitivity parameter. First, an inspection method and a defect detection algorithm are selected (not shown), and an arbitrary sensitivity parameter is set by a defect inspection apparatus (actual machine) (S11). According to the set sensitivity parameter, the inspection object is inspected for defects (S12). The operator reviews the result of the defect inspection offline with respect to the actual machine (S13), and determines the sensitivity table based on the sensitivity table created by the defect inspection apparatus (S14) (S15). If a desired detection level is achieved as a result of the review and the sensitivity table determination, the sensitivity parameter set in S11 is applied to the actual defect inspection. On the other hand, when the desired detection level is not achieved, the sensitivity parameter is changed and reset, and the processes from S11 to S15 are repeated.

Here, the sensitivity table refers to a visualization of the detection rate with respect to the number of inspections of actual measurement data obtained by performing the defect inspection twice or more.

FIG. 8 is a diagram showing the relationship between the response value of each detected program defect (value detected as a difference amount in the comparative inspection method) and the sensitivity parameter. For example, when the sensitivity parameter is set to an arbitrary high sensitivity threshold level 0, the detected real defects are r1, r2, and the nuisance defects are n1, n2, n3, n4, n5, n6, n7, n8. , so as to detect only the real defect r1 and r2, when shall set the sensitivity parameter, setting the sensitivity parameter at the level of the threshold value 1, of the nuisance defects, n4, n5, n6, n7 , N8 can be filtered so as to be excluded, but since the nuisance defects n1, n2, and n3 are detected in addition to the actual defects r1 and r2, the proper setting is not made.

  Next, when the threshold value 1 is changed to the threshold value 2 and reset, the detected defects are only r1 and r2, and the nuisance defects n1, n2, and n3 can also be filtered, so that the sensitivity parameter setting is optimized. It will be done.

  In FIG. 8, ten program defects have been described by way of example. Actually, however, the threshold value must be reset in order to optimize the sensitivity parameter from a large number of defects. In addition, when a plurality of defect detection algorithms are used, there are a plurality of parameters that need to be optimized, so that the number of repetitions of the threshold setting necessary for the optimization is further increased. Therefore, the time required for setting the sensitivity parameter shown in FIG. 7 depends on the skill level of the operator who performs the sensitivity parameter setting process and is inefficient.

Furthermore, as described above, even if the same sensitivity parameter is set, the detection rate may fluctuate each time a defect is detected due to various factors. Therefore, normally, the same sensitivity parameter is set multiple times (for example, 10 times). After detecting the defect and determining the detection rate relative to the number of detections, a sensitivity table is generated to check the detection rate, and the next sensitivity parameter is reset. Therefore, the occupation time of the actual machine is required for the above-mentioned multiple times of defect detection for each set sensitivity parameter, in addition to the time required for resetting the sensitivity parameter and the time required for generating the sensitivity table for each sensitivity parameter. Since time is added, the total time is enormous.

Therefore, conventionally, in order to optimize the sensitivity parameter in a short time, the sensitivity parameter is changed and the defect information is acquired a plurality of times, and the combination data in which each sensitivity parameter is associated with this defect information is acquired, Using this combination data, a function indicating the relationship between the desired statistical data and the sensitivity parameter is created, and the sensitivity parameter is determined based on this function, so that the sensitivity parameter can be optimized quantitatively and efficiently. It was possible to plan (for example, refer patent document 1).
JP-A-2004-138563 (claim 1)

  However, although the above-described conventional technique contributes to the reduction in the number of work steps described with reference to FIG. 7, it is necessary to acquire the defect information a plurality of times by changing the sensitivity parameter on the actual machine, and the expensive defect inspection apparatus (actual machine) is used. There is a disadvantage that the number of work steps is not reduced and the throughput of the apparatus is lowered. In addition, when multiple defect detection algorithms are used, the function for setting the sensitivity parameter becomes very complicated, and the number of combination data necessary to obtain the necessary statistics becomes enormous. There was an inconvenience that the number of man-hours for actual machine increases.

In view of the above problems, an object of the present invention is to provide a defect detection level adjustment method and a defect inspection system that reduce the number of man-hours for defect inspection using an actual machine and improve the throughput of the defect inspection apparatus.

In order to solve the above problems, the defect detection level adjusting method according to the present invention optimizes the sensitivity parameter set for the defect detection algorithm in order to set the detection level of the defect of the inspection object to a desired level. A defect detection level adjustment method, wherein the sensitivity parameter is set to an arbitrary high sensitivity by a defect inspection device, and a program defect created in the inspection object based on given specific data is detected. Obtaining actual detection data, and generating the review data from the actual detection data offline with respect to the defect inspection apparatus, and automatically generating a sensitivity table that visualizes the detection rate with respect to the number of inspections of the actual detection data , When the desired detection level is not achieved, the sensitivity parameter is reset on the offline, and the reset sensitivity parameter is reset. And acquisition of simulation detection data programmed defects due optimum to achieve the desired detection level by repeating the automatic generation of the simulation detection data sensitivity table a visualized detection rate of generation and the simulation detection data review data by The sensitivity parameter is set.

  According to this configuration, when the high sensitivity parameter is first set and the actual detection data is acquired by the defect inspection apparatus, the detection data when the sensitivity parameter is changed and reset is then used without using the actual machine. It can be acquired by simulation processing.

Also, the configuration of this, among the optimization efforts sensitivity parameter, the time required to generate the sensitivity table, it is possible to reduce the time required for the actual work.

The number of defects detected with respect to the set sensitivity parameter changes in a fixed relationship if the inspection object is the same and the defect inspection apparatus is stable. Therefore, from the actual detection data by the high sensitivity parameter, the relationship data of the change in the number of detected defects due to the change of the sensitivity parameter is obtained and stored for each defect detection algorithm, and when the sensitivity parameter is reset, the stored relationship Data may be read and the number of detected defects corresponding to the reset sensitivity parameter may be obtained.

In order to solve the above problems, the defect inspection system according to the present invention optimizes the sensitivity parameter set for the defect detection algorithm in order to achieve a desired level of defect detection of the inspection object, and detects the defect. A defect inspection system capable of adjusting a level, wherein a detection condition selection means capable of selecting one or more defect detection algorithms to be applied according to the property of the inspection object, and a given to the inspection object In order to detect a plurality of program defects created based on the specified data, parameter setting means for setting a sensitivity parameter for the selected defect detection algorithm, and defect detection using the set sensitivity parameter Defect detection processing means to be performed and obtained by this defect detection processing offline with respect to the defect detection processing means Simulation processing means for setting the optimum sensitivity parameter by simulation processing using the detection data, the simulation processing means comprising: review data generation means for generating review data from the actual detection data; and Sensitivity table generation means that visualizes the detection rate with respect to the number of inspections of actual detection data, sensitivity parameter change means that changes and resets the sensitivity parameter, and defect detection offline that acquires simulation detection data using the reset sensitivity parameter Processing means, and based on the simulation detection data, the review data generating means and the sensitivity table generating means generate review data and a sensitivity table.

  As is apparent from the above description, the defect detection level adjusting method and defect inspection system according to the present invention can reset the sensitivity parameter, estimate the number of detected defects based on the reset sensitivity parameter, and further review data. In addition, since the automatic generation of the sensitivity table can be performed by simulation processing, it is possible to reduce the man-hours for defect inspection using an actual machine and improve the throughput of the defect inspection apparatus.

  FIG. 1 is a block diagram of a defect inspection system according to the present invention. FIG. 1 illustrates the configuration necessary for the description of the present invention, and is not intended to limit the configuration. In the present embodiment, the die-to-database comparison method will be described as an example. However, the defect inspection method is not limited to this, and may be a die-to-die comparison method.

  Reference numeral 1 denotes a scanner that scans an object to be inspected such as a photomask (hereinafter referred to as “photomask or the like”) in at least one of a reflection illumination method and a transmission illumination method and reads a reflection image or a transmission image. . The scanner 1 may have a mode in which a laser light source is branched and a transmission inspection and a reflection inspection can be performed simultaneously.

  Of the arrows in FIG. 1, a one-dot chain line indicates a stage control signal, a two-dot chain line indicates actual image data, a solid line indicates design data, and a broken line indicates a flow of defect detection data.

An image such as a photomask read by the scanner 1 is transferred to the inspection processing system by the scanner control unit 2 in order to perform defect detection processing. The scanner control unit 2 also drives the stage (not shown) on which a photomask or the like is placed by the scanner 1 in the X, Y, and Z directions and controls the inclination θ.

  The data preparation processing unit 3 reads and collects pattern data from a pattern data server (not shown), and provides it to the system control and GUI processing unit 4 as inspection data. The system control / GUI processing unit 4 generates an image of design data in the development / reference image generation processing unit 5. In the data preparation processing unit 3 or system control / GUI processing unit 4, a defect detection algorithm to be applied is selected as an inspection condition in accordance with properties such as an inspection method and a photomask, and a sensitivity parameter is set.

  The actual image data such as a photomask transferred from the scanner 1 to the scanner control unit 2 and the design data generated by the development and reference image generation processing unit 5 are both sent to the distribution processing unit 6 so that they can be compared. Are distributed and transferred to the defect detection processing unit 7. The defect detection processing unit 7 compares the design data with the actual image data, detects a defective part where a difference occurs, and stores it together with the coordinate position data. For example, there is a comparison method in which the design data to be compared and the actual image data are converted into positive and negative electrical signals, and the comparison logic circuit adds both to detect a positive or negative signal as a defective part. .

  The detected defects are system controlled together with the above inspection conditions, the algorithm information for detecting each defect, and the reaction value of each defect that is the basis for determining the defect (value detected as a difference amount in the comparative inspection method). Are sent to the GUI processing unit 4 and stored as inspection result data. The stored inspection result data can be read out by the result management processing unit 8, output to a monitor and a printer, and stored in an external storage medium.

  By the way, the optimum sensitivity parameter for a photomask or the like can be set by detecting a program defect built in the photomask or the like. In the defect inspection system according to the present invention, the detection of the program defect is performed as described above. Instead of a series of processing from the scanner 1 to the result management processing unit 8 (that is, processing by an actual machine), simulation can be performed on the actual machine offline. Hereinafter, the configuration of the simulation processing unit 9 that optimizes the sensitivity parameter will be described. The program defect is created in the photomask or the like based on at least specific data composed of size data and coordinate data.

  Based on the inspection result data (hereinafter referred to as “actual detection data”) of defects detected by sensitivity parameters arbitrarily set for a photomask or the like in which program defects are created under the above inspection conditions The sensitivity table is generated by the review data generation unit 91 and the sensitivity table generation unit 92 of the simulation processing unit 9.

  The operator performs the review using the review data and the determination based on the sensitivity table, and when the desired detection level is not achieved, the sensitivity parameter changing unit 93 changes the arbitrarily set sensitivity parameter. Reset it.

  Based on the reset sensitivity parameter, the defect detection offline processing unit 94 simulates the defect detection process, that is, the process performed in the actual machine, and simulation detection data is obtained. Thereafter, the operation in the simulation processing unit 9 is repeated until the optimum sensitivity parameter that achieves the desired detection level can be specified.

  When the optimum sensitivity parameter can be specified, the sensitivity parameter can be set by the inspection condition of the actual machine, that is, the data preparation processing unit 3, and applied to the actual defect detection work.

  The simulation processing unit 9 may be configured as a dedicated device for setting sensitivity parameters in the defect inspection system, but a computer software program having the function of the simulation processing unit 9 is installed in a general-purpose personal computer, It is good also as a structure connected with the said real machine via LAN etc. so that communication is possible.

  FIG. 2 is a process flow diagram showing a procedure for optimizing sensitivity parameters according to the present invention. First, in consideration of the properties of the photomask etc. in which the program defect has been created, the optimum inspection method and defect detection algorithm are selected (not shown), and arbitrary sensitivity parameters are set by the defect inspection apparatus (actual machine) (S1).

Next, a defect inspection is performed using an actual machine such as a photomask with the set sensitivity parameter (S2). In addition, the defect inspection with this actual machine may be performed once, but in reality, the apparatus vibration of the defect detection device, the deterioration of the optical system, the light source fluctuation, the stage running accuracy (yawing and pitching), the stage stabilization accuracy, the atmospheric pressure Since the detection rate may fluctuate each time a defect is detected due to various factors such as fluctuations and temperature fluctuations, detection is usually repeated multiple times (for example, about 10 times for the same sensitivity parameter). Find the rate.

  From the result of the defect inspection by the actual machine, review data is generated offline for the actual machine and the review is performed (S3). Also, a sensitivity table is created from the result of the defect inspection by the actual machine (S4). This sensitivity table creation process can also be performed off-line.

  As a result of the review and sensitivity table determination (S5), if the desired detection level is achieved, defect inspection is performed with the sensitivity parameter set in S1. On the other hand, when the desired detection level is not achieved, the sensitivity parameter is changed and reset (S6), and the simulation inspection is performed off-line with the reset sensitivity parameter (S7). The processes of S3 to S7 are repeated until a desired detection level is reached.

  Compared with the conventional process flow diagram described with reference to FIG. 7, in the past, the sensitivity table creation, sensitivity resetting, and inspection routines after resetting were performed by the actual machine. In the present invention, these routines are offline. It can be performed by the simulation process.

  Note that the sensitivity parameter initially set on the actual machine is preferably set to a higher sensitivity than the so-called practical sensitivity. The above program defects are created in the inspected object based on specific data consisting of size data and coordinate data, but this is comprehensive with sensitivity parameters set to high sensitivity on the actual machine. This is because the detection process is performed and the specific data is stored, so that the resetting of the sensitivity parameter enables the simulation process based on the stored specific data.

  FIG. 3 is a diagram showing the relationship of the change in the number of detected defects accompanying the change in sensitivity parameter. In the inspection result, the inspection condition, the algorithm information for detecting each defect, and the reaction value of each defect (value detected as a difference amount in the comparative inspection method) are stored. The sensitivity parameter is a threshold value for determining a defect with respect to the reaction value of each defect. The number of defects detected for the set sensitivity parameter is constant for each defect detection algorithm if the same photomask or the like and the defect detection apparatus is stable. Therefore, when the reaction value is taken on the horizontal axis and the number of defects detected for each reaction value is taken on the vertical axis, a unique detection curve is obtained for each defect detection algorithm. Since a plurality of defect detection algorithms generally complement each other in order to detect defects of different types such as shapes and sizes, a certain defect may be detected by a plurality of defect detection algorithms.

  In FIG. 3, the detection curves of the defect detection algorithm A and the defect detection algorithm B are shown as an example. As described with reference to FIG. 2, the sensitivity parameter set arbitrarily at the beginning is set with high sensitivity, and if this detection curve is obtained from the inspection result, when the number of detection-defective lines is set, the line and the detection are detected. The optimum sensitivities Op1 and Op2 of the defect detection algorithm A and the defect detection algorithm B can be obtained near the point where the curves intersect. FIG. 3 illustrates an example in which one type of required defect detection number is determined. However, when the defect detection algorithm A and the defect detection algorithm B target defects having different characteristics, each detection curve In general, a different number of detected defects is set for each.

  Therefore, if the relationship shown in FIG. 3 is obtained, optimization of sensitivity parameter setting sensitivity can be quickly obtained, and filtering of the number of defects to be detected can be easily simulated.

  If the location to be detected as a defect is known in advance, the sensitivity parameter can be set for a specific defect instead of the number of defects. FIG. 8 is a diagram in which the reaction values of the detected program defects are plotted. For example, when the sensitivity parameter is set to an arbitrary high sensitivity threshold level 0, the detected real defects are r1, r2, and the nuisance defects are n1, n2, n3, n4, n5, n6, n7, n8. In order to set the sensitivity parameter so that only the actual defects r1 and r2 can be detected, the reaction values of the actual defects r1 and r2 may be confirmed, and the sensitivity parameter may be set to the threshold 2 level.

  FIG. 4 shows an example of the review data. In the present embodiment, the review data includes defect detection image data I1 and I2 and setting sensitivity parameter lists L1 and L2. In the defect detection image data I1 and I2, dots scattered at random are displayed as defects together with the pattern data image. The set sensitivity parameter lists L1 and L2 display the selected defect detection algorithm and sensitivity parameter values set for each defect detection algorithm for each illumination mode. In the present embodiment, defect detection algorithms ap1 to ap6 are selected for the transmission illumination method, and defect detection algorithms ar1 to ar3 are selected for the reflection illumination method.

  For example, in the defect detection image data I1, when the number of detected defects is large and it is determined that the number of detected defects is excessive, the number of detected defects may be filtered by reducing the sensitivity of some of the selected defect detection algorithms. . That is, in the present embodiment, the sensitivity parameter of the defect detection algorithm ap3 in the set sensitivity parameter list L1 is changed from 80 to 9, the sensitivity parameter of the defect detection algorithm ap4 is changed from 60 to 70, the sensitivity parameter 60 to 90 of ar2, and ar3 When the sensitivity parameter is changed from 80 to 90 (see the portion surrounded by the one-dot chain line in the set sensitivity parameter list L2), the white point detected as a defect is the defect detection image data as in the defect detection image data I2. It is possible to review the decrease compared to I1.

  FIG. 5 is a diagram showing an example of a sensitivity table. The sensitivity table is a visualization of the defect detection rate. Si1 in the sensitivity table is a column for displaying the type of program defect built into the photomask or the like. In this embodiment, twelve types from A to L are shown.

  Si2 indicates the number assigned to each program defect, and Si3 indicates the size (nm) of each program defect. Si4 indicates a desired defect level, and Si5 indicates an initial detection rate, that is, a detection rate based on actual detection data. On the other hand, Si6 resets the sensitivity parameter and shows the detection rate based on the simulation detection data.

  The numbers indicated by Si4 indicate the size of each program defect, and the numbers indicated by Si5 and Si6 indicate the detection rate of the size of the program defect indicated by Si4. Program defects corresponding to the portions shown by the Si4 and Si5 bar graphs show a 100% detection rate.

  Even if the defect detection process is performed under the same conditions, a fixed detection rate is not always obtained for various reasons. Therefore, normally, defect detection is performed a plurality of times (for example, 10 times), and the detection rate for these detection times is calculated.

  By automatically generating this sensitivity table on the off-line, the time required for generating the sensitivity table in the sensitivity parameter optimization work can be reduced from the time required for the actual machine work.

FIG. 6 is a diagram illustrating a method for acquiring reference detection data from actual defect data. In the description from FIG. 1 to FIG. 5, specific data such as size data and coordinate data of a program defect created in a photomask or the like is given in advance, but there is no such program defect (that is, specific data) For photomasks and the like (only actual defects whose data is unknown), it is not possible to generate review data or sensitivity tables by the above method.

  Therefore, the photomask and the like as described above are measured twice or more by the defect inspection apparatus (actual machine), and the reference detection data including the detection rate and the coordinate value of the detected defect is obtained. Then, the review data and sensitivity table are automatically generated off-line, and a desired defect detection level may be set from the generated review data and sensitivity table.

  FIG. 6 shows that six defects marked with ◯ are detected in the first inspection result, and seven defects marked with ▲ are detected in the second inspection result. When the total data of the two inspection results is calculated and displayed, the defect detected in the area surrounded by the dotted line is detected both in the first time and the second time, so the detection rate is 100%. Thus, the detection rate of each of three defects (one circle mark and two triangle marks) existing outside this region is 50%.

Accordingly, reference detection data is generated based on the detection rate and coordinate values of the detected defects, and review data and sensitivity tables are automatically generated off-line from the reference detection data, and the generated review data and sensitivity are generated. In order to obtain a desired defect detection level from the table, the sensitivity parameter may be optimized.

Block diagram of a defect inspection system according to the present invention Process flow diagram showing sensitivity parameter optimization procedure according to the present invention The figure which shows the relation of the change of the number of defect detection with the change of the sensitivity parameter Diagram showing data for review Diagram illustrating sensitivity table Diagram explaining how to obtain reference actual measurement data from actual defect data Process flow diagram showing conventional sensitivity parameter optimization procedure The figure which shows the relationship between the reaction value of each detected program defect, and a threshold value

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Scanner 2 Scanner control part 3 Data preparation process part 4 System control, GUI process part 5 Expansion | deployment, reference image generation process part 6 Distribution process part 7 Defect detection process part 8 Result management process part 9 Simulation process part

Claims (3)

A defect detection level adjusting method for optimizing a sensitivity parameter set for a defect detection algorithm in order to set a detection level of a defect of an inspection object to a desired level,
The sensitivity parameter is set to an arbitrary high sensitivity by a defect inspection device, a program defect created on the inspection object based on given specific data is detected to obtain actual detection data, and the defect A review table is generated off-line from the actual detection data offline with respect to the inspection device , and a sensitivity table visualizing the detection rate with respect to the number of inspections of the actual detection data is automatically generated to achieve the desired detection level. If not, the sensitivity parameter is reset on the offline, the simulation detection data of the program defect is acquired by the reset sensitivity parameter, the review data is generated by the simulation detection data, and the simulation detection data is the desired detection of the detection rate by repeating the automatic generation of visible sensitivity table Defect detection level adjustment method and sets the optimum sensitivity parameters to achieve the bell. From the actual detection data by the high sensitivity parameter, the relationship data of the change in the number of detected defects accompanying the change of the sensitivity parameter is obtained and stored for each defect detection algorithm, and when the sensitivity parameter is reset, the stored relationship data is 2. The defect detection level adjustment method according to claim 1, wherein the defect detection number corresponding to the reset sensitivity parameter is obtained by reading . A defect inspection system capable of optimizing a sensitivity parameter set for a defect detection algorithm and adjusting a defect detection level in order to set a defect detection level of an inspection object to a desired level.
Detection condition selection means capable of selecting one or more defect detection algorithms to be applied according to the properties of the inspection object, and a plurality of programs built on the inspection object based on given specific data In order to detect a defect, a parameter setting unit that sets a sensitivity parameter for the selected defect detection algorithm, a defect detection processing unit that performs defect detection using the set sensitivity parameter, and a defect detection processing unit On the other hand, it has simulation processing means for setting the optimum sensitivity parameter by simulation processing using the actual detection data obtained by this defect detection processing offline.
The simulation processing means includes a review data generating means for generating review data from the actual detection data, a sensitivity table generating means for visualizing a detection rate with respect to the number of inspections of the actual detection data, and changing the sensitivity parameter. Sensitivity parameter changing means for resetting and defect detection offline processing means for acquiring simulation detection data by the reset sensitivity parameter, and based on the simulation detection data, the review data generating means and the sensitivity table generating means A defect inspection system characterized by generating review data and a sensitivity table.