JP2006090901A - Treatment evaluation method and device in manufacturing process of semiconductor product, and semiconductor product manufacturing system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To evaluate the quality of a treatment executed in manufacture of a semiconductor product more accurately and quickly than hitherto without imposing a burden on an operator. <P>SOLUTION: A camera is installed beforehand so that the state wherein a processing treatment is executed to a substrate in a chamber is photographed, and an image when the processing treatment is executed without generating a failure among images acquired by photographing each state in the chamber when the processing treatment is executed to the substrate is stored beforehand as a model image in a storage means (#1). The state when the processing treatment is executed to the substrate in the chamber is photographed during manufacture of the semiconductor product (#3), and the quality of the processing treatment being executed in the chamber during manufacture of the semiconductor product is evaluated based on the image acquired in the step #3 and on the model image stored in the storage means (#4). <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a system, a method, and the like for performing an evaluation regarding a process performed when manufacturing a semiconductor product.

  The manufacturing process of a semiconductor product consists of many processing steps. Therefore, in order to improve the yield, it is more effective to verify whether each processing step is properly performed in real time during the manufacture of semiconductor products, rather than analyzing defective products and examining the cause of defects. Is.

  However, it is difficult for an operator to determine the presence or absence of a defect by observing the state of processing.

Therefore, a method as described in Patent Document 1 has been proposed. According to this method, the surface of the substrate to be processed when the wet etching process is performed is photographed, and an image obtained by photographing is monitored. Then, the in-plane etching distribution of the substrate to be processed is calculated according to the image. As a result, the worker can easily determine whether or not a defect has occurred, as compared with the case of simply observing the state of processing as in the conventional case.
JP 2004-60016 A

  However, according to the method described in Patent Document 1, in order to find a defective etching process, an operator must continuously monitor the change in the in-plane etching distribution. Although this monitoring work is easier than the case of directly observing the state of processing, it still places a heavy mental and physical burden on the worker. In addition, since monitoring largely relies on the experience and sense of the worker, it may overlook the occurrence of a defect or misrecognize that a defect has occurred even though no defect has occurred.

  It is also conceivable to record an image being processed and find out the occurrence of a defect later. However, this method slows down the detection of the occurrence of defects, and is not effective in improving the yield.

  In view of such problems, the present invention is capable of more accurately and promptly evaluating the quality of processing performed when manufacturing a semiconductor product than before without burdening workers. The purpose is to do.

  The process evaluation method according to the present invention is a process evaluation method in a semiconductor product manufacturing process for evaluating the quality of a process performed on a substrate on which the semiconductor product is based in order to manufacture the semiconductor product, A camera is set in advance so that the state when the processing is performed on the substrate is captured, and the state when the processing is performed on the substrate is captured by the camera. An image obtained when the processing is performed without any defect is stored in a storage unit as a model image in advance, and the processing is performed on the substrate during manufacture of the semiconductor product. Based on the first step of photographing with the camera, the image obtained in the first step, and the model image stored in the storage means. Te, and executes and a second step of evaluating the quality of the processing in the manufacturing of the semiconductor products.

  The processing evaluation method can be applied to a semiconductor manufacturing system. And the evaluation result by the said process evaluation method can be fed back in order to implement the process in a semiconductor manufacturing system more suitably. That is, when the process during the manufacture of the semiconductor product is evaluated as “defective” by the process evaluation method, the processing condition of the process is corrected and the processing apparatus that performs the process is controlled.

  According to the present invention, it is possible to accurately and promptly evaluate the quality of processing performed when manufacturing a semiconductor product without burdening workers.

  FIG. 1 is a diagram illustrating an overall configuration of a semiconductor product manufacturing system 100, FIG. 2 is a diagram illustrating an example of a process for manufacturing an IC chip, FIG. 3 is a diagram illustrating an example of a configuration of an etching apparatus 2F, and FIG. FIG. 5 is a diagram illustrating an example of a hardware configuration of the device 1, and FIG. 5 is a diagram illustrating an example of a functional configuration of the control device 1.

  The semiconductor product manufacturing system 100 is a CIM (Computer Integrated Manufacturing) that manufactures a semiconductor product such as an IC (Integrated Circuit) chip, a TFT array part of a liquid crystal display, an image sensor, or an amorphal solar cell, as shown in FIG. The control device 1, the production line ML, the communication line 3, and the like. The production line ML includes a plurality of processing devices 2 and a conveyor CY.

  A semiconductor product is manufactured by applying a number of processes to a wafer WF. For example, in the pre-process for manufacturing an IC chip, processes as shown in FIG. 2 are mainly performed in order. That is, first, a thin film 51 is formed on a wafer WF that has been subjected to a polishing process (round numeral 1). Photoresist 52 is applied to the formed wafer WF (circle numeral 2). The mask pattern of the circuit is transferred by exposing the wafer WF based on the design drawing prepared in advance (circle numeral 3).

  The photoresist 52 other than the exposed portion is removed (circle numeral 4). However, this is a case where a negative resist is used as the photoresist 52. When a positive resist is used, the exposed portion of the photoresist 52 is removed. The portion from which the photoresist 52 has been removed is etched to form a mask pattern on the wafer WF (circle numeral 5). The photoresist 52 remaining on the wafer WF is removed and removed (circle numeral 6). In addition, before and after these processes, a cleaning process, a drying process, a baking process, an impurity introduction process, and the like are appropriately performed as necessary.

  A circuit of the first layer of the IC chip is formed by a series of processing of the circled numbers 1 to 6. By repeating this series of processes as necessary, circuit layers are stacked to form an IC, and a pre-process for manufacturing an IC chip is completed.

  Semiconductor products other than IC chips are also manufactured by performing many types of processing on the wafer WF.

  Returning to FIG. 1, the manufacturing line ML is provided with various processing apparatuses 2 for performing the processes of the above-described steps. For example, as the processing apparatus 2, a P-CVD (Plasma Chemical Vapor Deposition) apparatus 2A for forming a thin film 51 on the wafer WF by chemical reaction by converting the gas in the chamber into plasma, and forming the thin film 51 on the wafer WF by sputtering. Sputtering apparatus 2B, photoresist coating apparatus 2C for applying a photoresist to wafer WF, stepper 2D for transferring a mask pattern to wafer WF, developing machine (developer) 2E for developing, etching apparatus 2F for etching, and photo A peeling device 2G for peeling the resist 52 from the wafer WF is provided.

  These processing devices 2 are connected to the control device 1 via the communication line 3 and perform processing to be performed on the wafer WF in accordance with a control command from the control device 1. The conveyor CY is also connected to the control device 1, and in accordance with a control command from the control device 1, the wafer WF that has been processed in a certain process is transported to the processing apparatus 2 for performing the process in the next process. Among the processes of these processes, the same processing apparatus 2 may be in charge of processes of a plurality of processes.

  Hereinafter, of the processes performed by these processing apparatuses 2, in particular, a case where the quality of the etching process and the drying process performed by the wet etching type and single wafer type etching apparatus 2 </ b> F is evaluated will be described as an example. . Hereinafter, wet etching is simply referred to as “etching”.

  As shown in FIG. 3, the etching apparatus 2F includes a controller 2F1, a communication apparatus 2F2, a first cleaning chamber CR1, an etching chamber CR2, a second cleaning chamber CR3, a first drying chamber CR4, a second drying chamber CR5, and the like. Is done.

  The first cleaning chamber CR1 cleans the wafer WF that has been transferred to the etching apparatus 2F. The etching chamber CR2 performs an etching process on the cleaned wafer WF. The second cleaning chamber CR3 cleans the wafer WF that has been subjected to the etching process. The first drying chamber CR4 removes water droplets on the surface of the wafer WF by rotating the cleaned wafer WF. That is, the wafer WF is spin-dried. The second drying chamber CR5 evaporates water droplets on the surface of the wafer WF by raising the temperature in the chamber. That is, the wafer WF is dried at a high temperature.

  Although the etching apparatus 2F can be configured to perform spin drying and high temperature drying in one chamber as one processing step, in this embodiment, spin drying and high temperature drying are performed in independent chambers. The etching apparatus 2F will be described as an example.

  In these five processing chambers (chambers), apparatuses for performing respective processing on the wafer WF are provided. Furthermore, a camera CM2 and a camera CM5 are installed in the etching chamber CR2 and the second drying chamber CR5, respectively. However, the installation position, shooting direction, shooting range, and the like of the camera CM2 are set so that it is possible to take a picture of the etching process performed on the wafer WF. The installation position, shooting direction, shooting range, and the like of the camera CM5 are set so that the wafer WF after being subjected to the high temperature drying process can be shot. As the camera CM2, an imaging device capable of continuous shooting such as a digital video camera is used. As for the camera CM5, a digital camera or the like is used because it is sufficient that the processed wafer WF can be photographed.

  The controller 2F1 controls each device provided in these five processing chambers, the communication device 2F2, and the cameras CM2 and CM5 based on a command received from the control device 1. The communication device 2F2 is used for communicating with the control device 1.

  Returning to FIG. 1, as shown in FIG. 4, the control device 1 includes a CPU 1a, a RAM 1b, a ROM 1c, a hard disk 1d, and various interfaces, and is used for controlling the processing device 2 and the conveyor CY. It is done.

  The hard disk 1d includes a program for realizing functions such as the processing device control unit 101, the image data receiving unit 102, the processing quality evaluation unit 103, the suitable recipe extraction unit 104, and the database management unit 105 as shown in FIG. Data is installed. These programs and data are loaded into the RAM 1b as necessary and executed by the CPU 1a. As the control device 1, a workstation or a personal computer is used.

  6 is a diagram showing an example of a database managed by the database management unit 105, FIG. 7 is a diagram showing an example of a recipe database RBf2, FIG. 8 is a diagram showing an example of a correction information database HB1, and FIG. 9 is a correction information database HB2. It is a figure which shows the example of.

  Next, the processing contents of each part of the control apparatus 1 shown in FIG. 5 and each part of the etching apparatus 2F shown in FIG. 3 will be described in detail.

  As shown in FIG. 6, the database management unit 105 in FIG. 5 stores and manages databases such as a recipe database RB, a correction information database HB (HB1, HB2,...), And a model image database MB.

  One recipe database RB is provided for each process performed by each processing apparatus 2. For example, in the case of the etching apparatus 2F that performs the five processes described in FIG. 3, recipe databases RBf1 to RBf5 are provided as recipe databases RB for the respective processes in the first cleaning chamber CR1 to the second drying chamber CR5. ing.

  “Recipe” means a production method, treatment method, blending method, or the like in the processing apparatus 2. That is, the temperature or degree of vacuum in the chamber, the type of gas to be used, the gas flow rate or temperature, the type of liquid to be used, the flow rate or temperature, the voltage to be pressurized, and the predetermined operation when the processing apparatus 2 performs processing. The setting conditions of various items related to the number of times or the speed of the power supply or the output frequency of the power source. The semiconductor product manufactured by the semiconductor product manufacturing system 100 is prepared with a recipe that indicates under what setting conditions it is preferable to perform processing for each processing. Hereinafter, such a suitable recipe is referred to as a “standard recipe” of the semiconductor product.

  Recipe information 7 indicating the contents of standard recipes of various semiconductor products manufactured by the semiconductor product manufacturing system 100 is stored in the recipe database RB. In addition to this, in order to cope with various changes in the environment, recipe information 7 of a large number of recipes in which the values of the respective setting conditions are variously stored is stored. For example, recipe information 72 as shown in FIG. 7 is stored in the recipe database RBf2 of the etching chamber CR2.

  In the recipe information 72, “recipe ID” is identification information for identifying the recipe information 72 from other recipe information 72. “Transfer speed” indicates the speed at which the wafer WF to be processed is transported into the chamber. The “swing speed” and “swing frequency” indicate the speed and number of times of swinging the wafer WF to be processed during the etching process, respectively. The “shower ejection flow rate” indicates the flow rate of the etching solution (chemical solution) fed into the chamber during the etching process. “Etching time” indicates the length of time for performing the etching process.

  In the correction information database HB1 of FIG. 6, correction data HDT is stored as shown in FIG. The correction data HDT is used to correct the recipe, that is, the processing conditions so that the defect is eliminated when a defect occurs during the etching process in the etching chamber CR2.

  “Bubble over number” and “absence / absence of wave abnormality” indicate the state of a defect that has occurred. The failure state is obtained by the processing quality evaluation unit 103 in FIG. This will be described later.

  The values in each field from “Shower ejection flow rate” to “Second cleaning room pure water cleaning time” are the values when a defect in the state shown in “Number of bubbles over” and “Whether there is a wave abnormality” occurs. This shows how to correct the processing conditions in each chamber in order to eliminate the defect. “Shower ejection flow rate” and “number of peristaltic movements” indicate the conditions of the flow rate of the etching liquid and the number of peristaltic times during the etching process in the etching chamber CR2. “First cleaning chamber pure water discharge amount” and “first cleaning chamber pure water cleaning time” indicate the flow rate and cleaning time of pure water during the cleaning process in the first cleaning chamber CR1, respectively. “Second cleaning chamber pure water discharge amount” and “second cleaning chamber pure water cleaning time” respectively indicate the flow rate and cleaning time of pure water during the cleaning process in the second cleaning chamber CR3.

  In the correction information database HB2 of FIG. 6, correction data HDU is stored as shown in FIG. In the correction data HDU, a wafer WF is a trace of water droplets (hereinafter referred to as “water droplet trace”) called a water spot or a water marker as a result of the high temperature drying process in the second drying chamber CR5. Is used to correct the recipe, that is, the processing conditions so that the defect disappears.

  The “water droplet trace over number” indicates the state of a defect that has occurred. The failure state is obtained by the processing quality evaluation unit 103 in FIG. This will be described later. The value of each field from “Second cleaning chamber pure water discharge amount” to “Pure water temperature” is used to eliminate the defects in the state shown in “Number of water droplets over”. It shows how the processing conditions in the first drying chamber CR4 should be corrected. The “second cleaning chamber pure water discharge amount” indicates the flow rate of pure water fed into the chamber during the cleaning process in the second cleaning chamber CR3. “Spin rotation speed” and “N2 temperature” respectively indicate the rotation speed when rotating the wafer WF to be processed and the temperature of the nitrogen gas in the chamber during the spin drying process in the first drying chamber CR4. Yes.

  The correction data HDT and HDU stored in the correction information databases HB1 and HB2 are obtained experimentally.

  Model image data 8M is stored in the model image database MB of FIG. The “model image” is a moving image obtained by taking an image of the inside of the chamber when the processing on the wafer WF is performed typically or ideally. The model image is referenced to evaluate the etching process failure. Therefore, the model image can also be called a “reference image”.

  The model image data 8M is prepared for each predetermined process in the process of manufacturing a semiconductor product. For example, for the etching apparatus 2F of this embodiment (see FIG. 3), model image data 8M for the etching process is prepared. However, even in the same type of processing in the same processing apparatus 2, if the recipe information 7 (that is, processing conditions) used when performing the processing is different, the state in the chamber being processed may be different. . Further, even if the same recipe information 7 is used, the state in the chamber being processed may be different if the processing target (that is, the semiconductor product to be manufactured) is different. Therefore, it is desirable to prepare the model image data 8M for each combination of the processing target and the recipe information 7 even if the model image data 8M is the same type of processing.

  The model image data 8M is prepared by the following procedure. For example, model image data of the etching process in the etching chamber CR2 of the etching apparatus 2F in the case where the semiconductor product X is an object to be manufactured and the conditions indicated in the recipe information 72 with “recipe ID = AZ2” are the processing conditions. 8M shall be prepared. First, a wafer WF is prepared in which the processing up to the previous etching process is performed for manufacturing the semiconductor product X.

  The etching process is started under the processing conditions indicated in the recipe information 72 with “recipe ID = AZ2”. At the same time, photographing of the state in the etching chamber CR2 is started by the camera CM2 provided in the etching chamber CR2. The photographing is continuously performed for a predetermined time (for example, until the etching process is completed). The shooting speed (number of frames taken per second) may be such that image analysis for evaluating the quality of the etching process described later can be performed. The shooting speed is set experimentally. When the shooting speed is about several frames per second, a digital camera may be used as the camera CM2.

  The quality of the etched wafer WF is inspected. If the quality exceeds a certain level, a moving image (image of each frame) obtained by photographing the state of the etching process this time is adopted as a model image. It is also possible to take an image several times and adopt the image when the quality is the highest. As will be described later, this model image is used to detect an abnormality related to the generation of waves on the water surface of the etchant during the etching process. Therefore, image processing for deleting the bubble image from each frame (frame) is performed.

  Then, model image data 8M is generated by converting the model image into data of a predetermined image format, and stored in the model image database MB. At this time, in order to distinguish from the other model image data 8M, the identifier of the semiconductor product X, the identifier of the used recipe information 7, the identifier of the main etching process, and the like are associated with each other.

  Further, the number of bubbles reflected in each frame of the model image data 8M is counted, and the count result is stored as model bubble data 8KH in association with the model image data 8M in the model image database MB.

  On the other hand, if the quality above a certain level is not obtained, the etching information and photographing described above are performed again until the quality above a certain level is obtained by changing the recipe information 72 to be used as necessary. Try again. In such a case, it is necessary to review the recipe information 72 used as the standard recipe.

  Returning to FIG. 5, the processing device control unit 101 controls each processing device 2 and the conveyor CY as follows. The conveyor CY is controlled so that the wafer WF that has undergone a certain process is transferred to the processing apparatus 2 or the chamber for the next process according to the process chart of the semiconductor product to be manufactured. For example, the wafer WF cleaned in the first cleaning chamber CR1 in FIG. 3 is transported to the next etching chamber CR2, and the wafer WF etched in the etching chamber CR2 is transferred to the next second cleaning chamber CR3. The conveyor CY is controlled so that the carrying operation is performed as follows. The processing device 2 is controlled by transmitting recipe information 7 for each processing together with a signal indicating a processing execution command to each processing device 2.

  For a while after the operation of the semiconductor product manufacturing system 100 is started, each processing apparatus 2 is controlled so as to be processed based on the recipe information 7 of the standard recipe. If any of the processes is found to be defective, the processing device 2 is controlled based on the recipe information 7 of the recipe that can eliminate the defect. That is, the processing apparatus 2 is controlled by correcting the processing conditions. This will be described later.

  In each processing apparatus 2, the process is performed on the wafer WF that has been subjected to the previous process in accordance with the command transmitted from the processing apparatus control unit 101 of the control apparatus 1. For example, in the etching apparatus 2F of FIG. 3, the controller 2F1 moves each apparatus in these chambers from the control apparatus 1 so that predetermined processing is performed in the first cleaning chamber CR1 or the second drying chamber CR5. Control is performed according to the five types of recipe information 7 transmitted.

  Further, the camera CM2 is controlled so as to photograph the state of the etching process performed in the etching chamber CR2 for a predetermined period. The start timing of imaging is set at the start of the etching process. That is, the timing is the same as the shooting timing for generating the model image data 8M described above. The shooting conditions such as the shooting speed, the shooting time, the shooting direction, and the shooting magnification are the same as those for shooting for generating the model image data 8M. Further, the camera CM 5 is controlled so as to photograph the surface of the wafer WF after the high temperature drying process is performed.

  The moving image data obtained by the camera CM2 is transmitted to the control device 1 by the communication device 2F2 as processing image data 8S. Image data obtained by the camera CM 5 is transmitted to the control device 1 as processed image data 8E.

  FIG. 10 is a flowchart illustrating an example of the flow of the etching pass / fail evaluation process, FIG. 11 is a diagram illustrating an example of a model image MG and a processing image SG of frames corresponding to each other, and FIG. FIG. 13 is a diagram for explaining an example of another method for detecting bubbles, FIG. 14 is a flowchart for explaining an example of the flow of water droplet trace quality evaluation processing, and FIG. 15 is a flow of suitable recipe extraction processing It is a flowchart explaining the example of.

  In the control device 1, the image data receiving unit 102 in FIG. 5 performs a process of receiving the processing-time image data 8S and the post-processing image data 8E transmitted from the etching device 2F.

  When the image data is received, the processing quality evaluation unit 103 performs processing in the etching chamber CR2 of the etching apparatus 2F based on the processing image data 8S and the model image data 8M stored in the model image database MB of FIG. The evaluation of the quality of the etching process performed is performed according to the procedure shown in FIG.

  First, the variable (x) indicating the frame number of the image to be analyzed is reset to “1”, and counters CT1 and CT2 for counting the number of abnormally generated bubbles and abnormally generated waves are counted. It is reset to “0” (# 101).

  The model image data 8M corresponding to the received processing image data 8S is extracted from the model image database MB (# 102). That is, the model image data 8M corresponding to the type of processing performed when the moving image of the processing-time image data 8S is photographed, the processing conditions (used recipe information 7), and the semiconductor product to be manufactured is extracted. To do. For example, for the purpose of manufacturing the semiconductor product X, the processing image data 8S when the etching process is performed in the etching chamber CR2 of the etching apparatus 2F under the processing conditions indicated by the recipe information 72 of “recipe ID = AZ2”. Is received, the model image data 8M associated with the identifier of the semiconductor product X, the identifier of the recipe information 72, and the identifier of the main etching process is extracted. As shown in FIG. 11, each frame of the moving image of the processing-time image data 8S and each frame of the extracted model image data 8M are arranged in time series. At this time, the frame at the start of photographing is defined as a first frame, and the subsequent frames are sequentially defined as a second frame, a third frame,..., An Nth frame. Hereinafter, an image shown in one frame of the processing time image data 8S is described as “processing time image SG”, and an image shown in one frame of the model image data 8M is described as “model image MG”. To do. In addition to the model image data 8M, model bubble data 8KH corresponding to the model image data 8M is also extracted.

  The number, position, and size of bubbles appearing in the processing image SG of the x-th frame (here, the first frame) of the processing image data 8S are detected (# 103). As a method for detecting these, for example, one of the following two methods is used.

  In the first method, as shown in FIG. 12A, the luminance values of pixels on one line continuous in the horizontal direction of the processing image SG are extracted, and the pixel values as shown in the graph of FIG. Find the horizontal change of. Then, it is determined that there is a bubble in a portion where the amount of change in luminance value between pixels at a predetermined interval is equal to or greater than a predetermined value. At this time, the position of the bubble is also recognized. Or you may discriminate | determine that there exists a bubble in the part whose luminance value is below a predetermined value. Such detection of the luminance value in the horizontal direction is repeated so as to sequentially scan downward from the upper end line of the processing-time image SG. Thereby, it is detected in which pixel in the processing image SG the bubble is reflected. Further, the bubble size can also be calculated based on the continuous number of pixels in which the bubble appears.

  Then, the detected bubbles are counted, and the number is stored together with the position and size of each bubble. However, a bubble whose size is equal to or smaller than a predetermined value has little effect on the quality of subsequent processing or is noise generated in the image sensor of the camera CM 2 (that is, the bubble is actually located at that position). Is not generated). Therefore, such bubbles are ignored.

  In the second method, only the bubble image region is extracted from the processing time image SG, and the bubble is detected. That is, as shown in FIG. 13, an image (standard image SGX) in which no bubbles are generated before the etching process is started is prepared in advance. The difference between the processing image SG and the standard image SGX is taken. Thereby, a differential image SGY in which only bubbles are shown is obtained. A black and white image SGZ represented by only “0” or “1” is obtained by binarizing the difference image SGY. Bubbles are detected by obtaining the position, size, and number of pixels having a pixel value of “1” from the monochrome image SGZ. As in the case of the first method, bubbles having a predetermined size or less are ignored.

  Returning to the flowchart, the number of bubbles in the xth frame (here, the first frame) indicated in the model bubble data 8KH extracted in step # 102 is compared with the number of bubbles detected in step # 103 ( # 104). Then, when a value obtained by adding a predetermined number to the former is used as a threshold value and the latter is equal to or greater than the threshold value (Yes in # 104), it is determined that there is an abnormality in the generation of bubbles in the scene of the frame, and the counter “1” is added to CT1 (# 105). If it is less than the threshold (No in # 104), it is determined that there is no abnormality and no addition is performed.

  Based on the position and size of the bubbles detected in step # 103, in order to more accurately compare the generation of waves in the model image MG and the generation of waves in the processing-time image SG in step # 107. The bubble image is deleted from the processing image SG (# 106).

  The processing-time image SG from which the bubbles are deleted and the model image MG having the same frame number (here, the first frame) are matched to detect the difference in the generation of the waves (# 107). For example, the two images are overlapped so that the pixel representing the upper end of the leftmost wave of the processing image SG and the processing image SG overlap. The overlapping state of waves in the right direction from this pair of waves is detected. If a pair of waves with a pixel shift equal to or greater than a predetermined interval is detected (Yes in # 108), it is determined that there is an abnormality in the generation of waves in the scene of the frame, and the counter CT2 is set to “1”. Are added (# 109). If not detected (No in # 108), it is determined that there is no abnormality, and the addition to the counter CT2 is not performed.

  The comparison target is changed to the model image MG and the processing image SG of the next frame (# 110), and the determination of the quality of the generation of bubbles and waves is repeated until the last frame (Nth frame) (# 103- No in # 109 and # 111).

  When the determination for the last frame (Nth frame) is completed (Yes in # 111), the etching when the moving image of the processing-time image data 8S is taken based on the values of the counters CT1 and CT2 It is evaluated whether or not the processing is good (# 112). In the present embodiment, the values of the counters CT1 and CT2 are compared with the respective threshold values, and when either of them is equal to or greater than the threshold value (Yes in # 112), the current etching process is not good, that is, an abnormality has occurred. (# 114). If both are less than the threshold (No in # 112), it is determined that the current etching process was good (# 113).

  For example, when the threshold value of the counter CT1 is set to “one third of the number of frames” and the threshold value of the counter CT2 is set to “one half of the number of frames”, “CT1 ≧ N / 3 or CT2 ≧ N / 2 ”, this etching process is evaluated as not good.

  In parallel with the processing of FIG. 10, the processing quality evaluation unit 103 evaluates the quality of the remaining water droplet trace based on the post-processing image data 8E transmitted from the etching apparatus 2F, for example, as shown in FIG. Follow the procedure below.

  An image of the surface of the wafer WF having no water droplet trace prepared in advance and an image of the surface of the wafer WF after the high temperature drying process shown in the processed image data 8E are superimposed (# 121 in FIG. 14). By obtaining the difference between the two images, an image of only the water droplet trace is generated (# 122). The water droplet traces reflected in the image are counted (# 123). If the number is greater than or equal to the threshold value (Yes in # 124), it is evaluated that the water droplet trace is defective because it remains on the surface of the wafer WF (# 126). If it is less than the threshold value (No in # 124), it is evaluated as good (# 125).

  Returning to FIG. 5, when the processing quality evaluation unit 103 evaluates that the defect has occurred in the processing of any of the processing apparatuses 2, the suitable recipe extraction unit 104 eliminates the defect and obtains a preferable processing result. The processing condition for making it possible, that is, the recipe information 7 is discriminated and extracted by the procedure as shown in FIG.

  For example, if the processing quality evaluation unit 103 evaluates that a defect has occurred in the etching process of the etching apparatus 2F, the suitable recipe extraction unit 104 first obtains the state of the etching process failure (# 131). . That is, it is determined whether or not there is an abnormality in how the waves are generated and how many bubbles are generated above the threshold. The former is obtained from the value of the counter CT2 finally obtained by the processing of FIG. For the latter, for example, an average value of the difference between the number of bubbles detected in step # 103 and the threshold value may be calculated for each frame.

  Correction data HDT corresponding to the obtained defective state is extracted from the correction information database HB1 of FIG. 8 (# 132). For example, if the difference between the generated bubble and the threshold is 7 and there is a wave abnormality, “bubble over number = 5 to 10” and “wave presence / absence = present” The correction data HDT is extracted.

  Based on the extracted correction data HDT, recipe information 7 of the following recipe is extracted from the recipe databases RBf1 to RBf3 (# 133).

  Shown from the recipe database RBf2, the shower ejection flow rate and the number of perturbations are extracted from the correction data HDT, which are the values indicated in the “shower ejection flow rate” and the “number of peristaltic motions”, and the conditions (conveying speed, peristaltic motion) of other items. The recipe information 7 of the recipe having the same value as the current processing condition is extracted. That is, the recipe information 7 that can change the current processing condition to the values shown in the “shower ejection flow rate” and the “number of peristaltic movements” of the correction data HDT is extracted.

  For example, the current processing condition is the condition indicated in the recipe information 7 (see FIG. 7) of “recipe ID = AZ2”, and the correction data HDT extracted in step # 132 is “bubble count = 5 to 10”. ”And“ the presence / absence of wave abnormality = present ”. In this case, it is necessary to correct the “shower ejection flow rate” and the “number of peristaltic movements” to values in the ranges of “135 to 180” and “3 to 4”, respectively. In addition, it is necessary to keep the processing conditions of the other items as shown in the recipe information 7 of “recipe ID = AZ2”. Therefore, the recipe information 7 of “recipe ID = AXA” that satisfies these requirements is extracted.

  Similarly, from the recipe database RBf1, the pure water discharge amount and the pure water cleaning time are values indicated by “first cleaning chamber pure water discharge amount” and “first cleaning chamber pure water cleaning time” of the correction data HDT, respectively. Yes, and the recipe value 7 of the recipe, in which the value of the condition of the other items is the same as the value of the current processing condition, is extracted. This recipe information 7 is used for the cleaning process in the first cleaning chamber CR1. From the recipe database RBf3, the pure water discharge amount and the pure water cleaning time are values indicated by “second cleaning chamber pure water discharge amount” and “second cleaning chamber pure water cleaning time” in the correction data HDT, respectively, and The recipe information 7 of the recipe in which the value of the condition of the other items is the same as the value of the current processing condition is extracted. This recipe information 7 is used for the cleaning process in the second cleaning chamber CR3.

  Further, even when the processing quality evaluation unit 103 evaluates that a defect related to the water droplet trace remaining on the surface of the wafer WF has occurred, the recipe information 7 is extracted in the same procedure as in the case of the etching processing failure. That is, first, the state of the generated defect is obtained (# 131). Here, the difference between the water droplet trace remaining on the wafer WF and the threshold value (hereinafter referred to as the “water droplet trace over number”) is obtained. From the correction information database HB2 (see FIG. 9), correction data HDU corresponding to the number of water drop traces is extracted (# 132).

  Then, based on the extracted correction data HDU, recipe information 7 of the following recipe is extracted from the recipe databases RBf3 and RBf4 (# 133). Recipe information 7 is extracted from the recipe database RBf3 so that the pure water discharge amount can be changed to the value of the “second cleaning chamber pure water discharge amount” of the correction data HDU. This recipe information 7 is used for the cleaning process in the second cleaning chamber CR3. Recipe information 7 that can change the spin rotation speed, N2 temperature, and pure water temperature from the recipe database RBf4 to the values of “spin rotation speed”, “N2 temperature”, and “pure water temperature” of the correction data HDU, respectively. Extract. This recipe information 7 is used for the spin drying process in the first drying chamber CR4.

  Returning to FIG. 5, the recipe information 7 extracted by the preferred recipe extraction unit 104 is used for controlling the processing apparatus 2 such as the etching apparatus 2 </ b> F after the occurrence of a defect is detected. That is, suitable recipe information 7 for eliminating defects is fed back for control correction.

  FIG. 16 is a flowchart for explaining an example of the overall processing flow of the semiconductor product manufacturing system 100. Next, a processing flow of the semiconductor product manufacturing system 100 when manufacturing a certain semiconductor product will be described with reference to a flowchart of FIG.

  The state of the etching process when no defect occurs is photographed in advance to generate model image data 8M (# 1). The operation of the semiconductor product manufacturing system 100 is started (# 2). At this time, control of each processing apparatus 2 is started based on the standard recipe corresponding to the semiconductor product to be manufactured.

  The processing performed in the etching chamber CR2 of the etching apparatus 2F and the wafer WF after the processing in the second drying chamber CR5 are photographed (# 3).

  The quality of the etching process is determined by comparing the image of each frame of the moving image data (processed image data 8S) obtained by shooting with the image of each frame of the model image data 8M prepared in advance. Evaluate (# 4). The evaluation procedure is as described above with reference to FIG. In parallel with this, the quality of the water droplet trace remaining on the wafer WF is evaluated based on the processed image data 8E (# 5). The evaluation procedure is as described above with reference to FIG.

  If it is evaluated that there is a defect (Yes in # 6), a recipe suitable for correcting it is extracted (# 7). The extraction procedure is as described above with reference to FIG. Then, each processing device 2 is controlled based on the extracted recipe (# 8). That is, a suitable recipe is fed back for correction of processing conditions.

  The processes of steps # 3 to # 8 are repeatedly performed until the target semiconductor product is manufactured (No in # 9).

  According to the present embodiment, an evaluation of the quality of the processing performed at the time of manufacturing a semiconductor product is performed by using an image of a state of good processing prepared in advance and an image of processing when actually manufacturing. Do by comparing. Thereby, it can perform more correctly and rapidly than before, without putting a burden on an operator.

  Processing-time image data 8S, post-processing image data 8E, data relating to bubbles and waveforms generated in the etching solution during etching processing, and data relating to water droplet traces remaining on the wafer WF after high-temperature drying processing are accumulated as history data. Also good. The developer of the semiconductor product manufacturing system 100 can use the history data for analysis and investigation of the tendency of occurrence of defects (data mining) and the like to be used for improving the semiconductor product manufacturing system 100.

  It is desirable to evaluate processing on the wafer WF such as etching processing and high-temperature drying processing in real time as soon as the processing image data 8S and the processed image data 8E are input to the control device 1. However, if it cannot be executed immediately, these image data may be temporarily swapped to the hard disk 1d (see FIG. 4) and loaded into the RAM 1b when it becomes executable.

  In the present embodiment, the model image data 8M, the processing image data 8S, the processed image data 8E, and the like are input to the control device 1 via the communication line 3, but via a removable disk such as a CD-R or MO. It is also possible to input.

  In the present embodiment, the etching process and the high-temperature drying process in the etching apparatus 2F are evaluated, but the present invention can also be applied to the evaluation of other processes. For example, the spin drying process performed in the first drying chamber CR4 (see FIG. 3) of the etching apparatus 2F can be evaluated. In this case, the wafer WF after the spin drying process may be photographed with a camera, and water droplets remaining on the wafer WF may be detected from the image and evaluated based on the number thereof. The present invention can also be applied to processing performed in the processing apparatus 2 other than the etching apparatus 2F.

  FIG. 17 is a diagram illustrating an example of the defect case image data 8B. In the present embodiment, the evaluation of the high-temperature drying process is performed based on the number of water droplet traces remaining on the wafer WF, but may be performed based on the area of the water droplet traces. Alternatively, the evaluation of the etching process was performed by comparing the processing-time image data 8S with the model image data 8M. The case image data 8B) may be prepared as reference image data, and this may be compared with the processing image data 8S. By comparing with the defect case image data 8B, the cause of the defect can be easily determined. For example, if it is detected that an image that substantially matches the image BGa is included in the processing-time image data 8S, it is considered that the cause of the defect is that the carry-in speed into the etching chamber CR2 is too high. If an image substantially coincident with the image BGb is detected, the bias in the shower is considered to be the cause of the failure. If an image that substantially matches the image BGc is detected, it is considered that the bubbling abnormality is the cause of the failure.

  In addition, the configuration of the entire semiconductor product manufacturing system 100, the control device 1, the manufacturing line ML, and the processing device 2 or each part, the processing content, the processing order, and the like can be appropriately changed in accordance with the spirit of the present invention.

The following additional notes are disclosed with respect to the embodiment described above.
(Appendix 1) A process evaluation method in a manufacturing process of a semiconductor product, which evaluates the quality of a process performed on a substrate on which the semiconductor product is based in order to manufacture the semiconductor product,
A camera is installed in advance so that the state when the processing is performed on the substrate is photographed,
Of the images obtained by photographing the state when the processing is being performed on the substrate with the camera, an image when the processing is performed without causing a defect is used as a model image. Store in the storage means in advance,
A first step of photographing with the camera a state when the processing is being performed on the substrate during the manufacture of the semiconductor product;
A second step of evaluating the quality of the processing during manufacture of the semiconductor product based on the image obtained in the first step and the model image stored in the storage unit;
The process evaluation method in the manufacturing process of the semiconductor product characterized by performing these.
(Additional remark 2) It is the process evaluation apparatus in the manufacturing process of a semiconductor product which evaluates the quality of the process implemented with respect to the board | substrate used as the basis of the said semiconductor product in order to manufacture a semiconductor product,
An image input means for inputting an image taken by a camera installed so as to photograph the state when the processing is performed on the substrate;
Of the images obtained by photographing the state when the processing is being performed on the substrate with the camera, an image when the processing is performed without causing a defect is used as a model image. Storage means for storing;
When an image at the time of manufacture, which is an image taken when the processing is performed on the substrate, taken by the camera during the manufacture of the semiconductor product is input by the image input means, it is input. A process evaluation unit that evaluates the quality of the process during the manufacture of the semiconductor product based on the manufacturing-time image and the model image stored in the storage unit;
A processing evaluation apparatus in a manufacturing process of a semiconductor product, characterized by comprising:
(Additional remark 3) The said process evaluation means calculates | requires the difference of the said model image and the said image at the time of manufacture, and evaluates the quality of the said process based on the said difference,
The processing evaluation apparatus according to attachment 2.
(Additional remark 4) The said memory | storage means memorize | stores the said model image of several continuous frames of the predetermined time slot | zone in the period when the said process is implemented,
The image input means inputs the production-time images of a plurality of consecutive frames in the predetermined time period,
The processing evaluation means obtains a difference between the model image and the manufacturing image of frames corresponding to each other in time, counts the number of frames in which the difference is a predetermined value or more, and the counted number is small Evaluate that the treatment is good,
The processing evaluation apparatus according to attachment 2.
(Additional remark 5) The said process is an etching process,
The processing evaluation means evaluates the quality of the processing based on a difference in the state of the wave or bubbles on the water surface of the etching solution shown in each of the model image and the production-time image,
The processing evaluation apparatus according to Supplementary Note 3 or Supplementary Note 4.
(Additional remark 6) It is the process evaluation apparatus in the manufacturing process of a semiconductor product which evaluates the quality of the drying process implemented with respect to the board | substrate used as the basis of the said semiconductor product in order to manufacture a semiconductor product,
Image input means for inputting an image showing the substrate after the drying process is performed;
Water drop counting means for counting water drop traces reflected in the input image;
A process evaluation means for evaluating the quality of the drying process based on the counted number of water droplet traces;
A processing evaluation apparatus in a manufacturing process of a semiconductor product, characterized by comprising:
(Appendix 7) A semiconductor product manufacturing system for manufacturing a semiconductor product by processing a substrate,
An image input means for inputting an image photographed by a camera installed so that a state of performing a predetermined process on the substrate is photographed;
Of the images obtained by photographing with the camera the state when the predetermined process is being performed on the substrate, the image when the predetermined process is performed without causing a defect Storage means for storing as a model image;
When an image at the time of manufacture, which is an image taken when the predetermined processing is performed on the substrate, photographed by the camera during the manufacture of the semiconductor product is input by the image input unit, A process evaluation unit that evaluates the quality of the predetermined process during manufacture of the semiconductor product based on the input image at the time of manufacture and the model image stored in the storage unit;
A processing condition correction unit that corrects a processing condition of the predetermined processing when the processing evaluation unit evaluates that the predetermined processing during manufacture of the semiconductor product is defective;
A semiconductor product manufacturing system comprising:
(Appendix 8) A semiconductor product manufacturing system for manufacturing a semiconductor product by processing a substrate,
A processing apparatus for performing predetermined processing on the substrate;
A camera installed so as to photograph the state when the predetermined processing is performed;
A control device for controlling the processing device,
In the control device,
Of the images obtained by photographing with the camera the state when the predetermined process is being performed on the substrate, the image when the predetermined process is performed without causing a defect Storage means for storing as a model image;
An image input means for inputting, as a production-time image, an image of the state when the predetermined processing is performed on the substrate, which is taken by the camera during the manufacture of the semiconductor product;
When the manufacturing-time image is input, the processing evaluation is performed to evaluate the quality of the predetermined processing during the manufacturing of the semiconductor product based on the manufacturing-time image and the model image stored in the storage unit. Means,
A processing condition correction unit that corrects a processing condition of the predetermined processing when the processing evaluation unit evaluates that the predetermined processing during manufacture of the semiconductor product is defective;
Is provided,
When the processing condition of the predetermined processing is corrected, the control device controls the processing device based on the corrected processing condition.
A semiconductor product manufacturing system characterized by that.
(Supplementary note 9) A computer program used in a computer for evaluating the quality of processing performed on a substrate on which a semiconductor product is based to manufacture a semiconductor product,
Of the images obtained by photographing the processing when the processing is being performed on the substrate, an image when the processing is performed without causing a defect is stored as a model image. Processing to access the storage means,
During the manufacture of the semiconductor product, a process of inputting an image obtained by photographing the state when the processing is performed on the substrate;
Based on the input image and the model image stored in the storage means, a process for evaluating the quality of the processing during the manufacturing of the semiconductor product,
A computer program for causing a computer to execute.
(Additional remark 10) It is the process evaluation method in the manufacturing process of a semiconductor product which evaluates about the process performed with respect to the board | substrate used as the basis of the said semiconductor product in order to manufacture a semiconductor product,
A camera is installed in advance so that the state when the processing is performed on the substrate is photographed,
Of the images obtained by previously capturing the state when the processing is performed on the substrate with the camera, an image when the processing is performed when a defect occurs is referred to as a reference image. As previously stored in the storage means,
A first step of photographing with the camera a state when the processing is being performed on the substrate during the manufacture of the semiconductor product;
A second step of evaluating whether or not the processing during the manufacture of the semiconductor product is defective based on the image obtained in the first step and the reference image stored in the storage means. When,
The process evaluation method in the manufacturing process of the semiconductor product characterized by performing these.

  INDUSTRIAL APPLICABILITY The present invention is preferably used for improving the yield of semiconductor products as compared with the conventional one without placing a burden on workers and providing inexpensive and high-quality semiconductor products.

It is a figure which shows the whole structure of a semiconductor product manufacturing system. It is a figure which shows the example of the process of manufacturing an IC chip. It is a figure which shows the example of a structure of an etching apparatus. It is a figure which shows the example of the hardware constitutions of a control apparatus. It is a figure which shows the example of a functional structure of a control apparatus. It is a figure which shows the example of the database etc. which a database management part manages. It is a figure which shows the example of a recipe database. It is a figure which shows the example of a correction information database. It is a figure which shows the example of a correction information database. It is a flowchart explaining the example of the flow of an etching quality evaluation process. It is a figure which shows the example of the model image of a mutually corresponding flame | frame, and the image at the time of a process. It is a figure for demonstrating the example of the method of detecting a bubble. It is a figure for demonstrating the example of the other method of detecting a bubble. It is a flowchart explaining the example of the flow of a water drop trace quality evaluation process. It is a flowchart explaining the example of the flow of a suitable recipe extraction process. It is a flowchart explaining the example of the flow of the whole process of a semiconductor product manufacturing system. It is a figure which shows the example of defect example image data.

Explanation of symbols

100 Semiconductor Product Manufacturing System 1 Control Device (Processing Evaluation Device)
102 Image data receiving unit (image input means)
103 Processing quality evaluation unit (processing evaluation means)
104 preferred recipe extraction unit (processing condition correction means)
CM2 camera MB model image database (storage means)
MG Model image WF Wafer (Substrate)
SG processing image (manufacturing image)

Claims (3)

A process evaluation method in a manufacturing process of a semiconductor product, which evaluates the quality of a process performed on a substrate on which the semiconductor product is based in order to manufacture a semiconductor product,
A camera is installed in advance so that the state when the processing is performed on the substrate is photographed,
Of the images obtained by photographing the state when the processing is being performed on the substrate with the camera, an image when the processing is performed without causing a defect is used as a model image. Store in the storage means in advance,
A first step of photographing with the camera a state when the processing is being performed on the substrate during the manufacture of the semiconductor product;
A second step of evaluating the quality of the processing during manufacture of the semiconductor product based on the image obtained in the first step and the model image stored in the storage unit;
The process evaluation method in the manufacturing process of the semiconductor product characterized by performing these. A process evaluation apparatus in a manufacturing process of a semiconductor product that evaluates the quality of a process performed on a substrate that is the basis of the semiconductor product to manufacture a semiconductor product,
An image input means for inputting an image taken by a camera installed so as to photograph the state when the processing is performed on the substrate;
Of the images obtained by photographing the state when the processing is being performed on the substrate with the camera, an image when the processing is performed without causing a defect is used as a model image. Storage means for storing;
When an image at the time of manufacture, which is an image taken when the processing is performed on the substrate, taken by the camera during the manufacture of the semiconductor product is input by the image input means, it is input. A process evaluation unit that evaluates the quality of the process during the manufacture of the semiconductor product based on the manufacturing-time image and the model image stored in the storage unit;
A processing evaluation apparatus in a manufacturing process of a semiconductor product, characterized by comprising: A semiconductor product manufacturing system for manufacturing a semiconductor product by processing a substrate,
An image input means for inputting an image photographed by a camera installed so that a state of performing a predetermined process on the substrate is photographed;
Of the images obtained by photographing with the camera the state when the predetermined process is being performed on the substrate, the image when the predetermined process is performed without causing a defect Storage means for storing as a model image;
When an image at the time of manufacture, which is an image taken when the predetermined processing is performed on the substrate, photographed by the camera during the manufacture of the semiconductor product is input by the image input unit, A process evaluation unit that evaluates the quality of the predetermined process during manufacture of the semiconductor product based on the input image at the time of manufacture and the model image stored in the storage unit;
A processing condition correction unit that corrects a processing condition of the predetermined processing when the processing evaluation unit evaluates that the predetermined processing during manufacture of the semiconductor product is defective;
A semiconductor product manufacturing system comprising:

Images