Automatic quality control method and device for production line, and storage medium

Description: TECHNICAL FIELD [Detailed description of the invention]

【0001】

[Technical field to which the invention belongs]

The present invention relates to a quality control method for a production line that automatically controls the quality of the production line for various products such as a semiconductor production line, an apparatus thereof, and a storage medium.

【0002】

[Conventional technology]

Product quality control in the production line of various products is performed by sampling the data of a small number of products from a large number of products to be manufactured, and depending on the overall distribution and the degree of variation of the sampled small number of products, the manufacturing conditions and the manufacturing conditions. It is done by understanding the relationship between the quality of the product. In addition, quality control such as product defect analysis using more detailed data than these sampled data has gained many years of experience on the manufacturing line, and workers who have come to have unique rules of thumb and improvement skills. In many cases it is possible only by.

【0003】

Nowadays, technologies such as monitoring networks and databases have been developed so that it is relatively easy to manage and store manufacturing conditions and performance information of all products that flow on the manufacturing line by bar code number etc. Is becoming.

【0004】

[Problems to be Solved by the Invention]

However, although it seems that more advanced quality control will be possible if a huge amount of manufacturing history data is effectively used, human data processing capacity is limited. In addition, defect analysis using detailed data often relies on the judgment and intuition of humans with unique empirical rules and improvement skills, which further hinders the effective use of data. That is, it is difficult to perform quality control by effectively utilizing a large amount of manufacturing history information by the conventional method depending on the human system.

【0005】

Therefore, an object of the present invention is to provide an automatic quality control method for a production line, an apparatus thereof, and a storage medium that can effectively utilize a large amount of data by overcoming ambiguity that relies on human data processing ability, empirical rules, and intuition. And.

【0006】

[Means for solving problems]

The present invention according to claim 1 monitors manufacturing condition data and product workmanship data on the manufacturing line and accumulates them in a manufacturing history database, of which the workmanship data is monitored to detect quality inhibition events on the manufacturing line. Then, the quality-inhibiting factor is extracted in consideration of the manufacturing condition data, the extraction result is compared with the quality improvement history data stored in advance to confirm the validity, and the manufacturing line is further checked for the quality-inhibiting factor. It is characterized by executing a simulation of the phenomenon related to the above, verifying the validity, and if there is such validity and validity, changing the manufacturing conditions so as to improve the quality inhibitory factor and feeding back to the manufacturing line. This is an automatic quality control method for the production line.

【0007】

According to the second aspect of the present invention, in the automatic quality control method of the manufacturing line according to the first aspect, when the validity and the validity cannot be confirmed by the collation with the quality improvement history data and the simulation, the present invention is described. It is characterized by presenting a warning and the candidate for the quality inhibitory factor and asking whether or not to change the manufacturing conditions in order to improve the quality inhibitory factor.

【0008】

The present invention according to claim 3 is characterized in that, when the quality inhibiting factor is improved in the automatic quality control method for the production line according to claim 1, this improvement case is added to the quality improvement history data. To do.

【0009】

The present invention according to claim 4 is a monitoring data conversion means for monitoring manufacturing condition data and product workmanship data in a manufacturing line and accumulating them in a manufacturing history database, and monitoring the workmanship data stored in the manufacturing history database. When a quality inhibition event on the production line is detected, the analysis means for extracting the quality inhibition factor in consideration of the production condition data is collated with the extraction result and the quality improvement history data stored in advance. Validity confirmation means for confirming the validity, validity verification means for verifying the validity by executing a simulation of a phenomenon related to the manufacturing line for the quality-inhibiting factor, and if there is such validity and validity, The automatic quality control device for a production line is provided with a feedback means for changing production conditions and feeding back to the production line so as to improve the quality-inhibiting factor.

【0010】

The present invention according to claim 5 warns when the validity and validity cannot be confirmed by the collation with the quality improvement history data and the simulation in the automatic quality control device of the manufacturing line according to claim 4. It is characterized in that it is provided with a warning determination means for presenting the quality inhibitory factor candidate and asking whether or not to change the manufacturing condition in order to improve the quality inhibitory factor.

【0011】

The present invention according to claim 6 is provided with a learning means for adding the improvement case to the quality improvement history data when the quality inhibiting factor is improved in the automatic quality control device of the manufacturing line according to claim 4. It is characterized by that.

【0012】

According to claim 7, the present invention monitors manufacturing condition data and product workmanship data on a manufacturing line and stores them in a manufacturing history database, of which the workmanship data is monitored to detect quality inhibition events on the manufacturing line. Then, the quality-inhibiting factor is extracted by taking the manufacturing condition data into consideration, and the extraction result is collated with the quality improvement history data stored in advance to confirm its validity, and further, the manufacturing line regarding the quality-inhibiting factor. A program is stored in which a simulation of a phenomenon related to the above is executed to verify the validity, and if there is such validity and validity, the manufacturing conditions are changed so as to improve the quality-inhibiting factor and the data is fed back to the manufacturing line. It is a storage medium characterized by the fact that it has been done.

【0013】

The present invention according to claim 8 warns the storage medium according to claim 7 when the validity and validity cannot be confirmed by collation with the quality improvement history data and the simulation. It is characterized in that it presents candidates and asks whether or not to change the manufacturing conditions in order to improve the quality-inhibiting factor.

【0014】

The present invention according to claim 9 is characterized in that, when the quality-inhibiting factor is improved in the storage medium according to claim 7, this improvement case is added to the quality improvement history data.

【0015】

BEST MODE FOR CARRYING OUT THE INVENTION

(1) Hereinafter, the first embodiment of the present invention will be described with reference to the drawings.

【0016】

FIG. 1 is a block configuration diagram of an automatic quality control device on a manufacturing line. The manufacturing process (manufacturing line) 1 to be controlled is, for example, a semiconductor manufacturing process or a color brown tube manufacturing process, and these steps include film formation, resist coating, exposure, development, etching, and resist removal. The manufacturing condition control unit 2 of the manufacturing process 1 has a function of controlling the manufacturing conditions in the manufacturing process 1, such as temperature, humidity, resist viscosity, resist coating strength, resist coating amount, and the like.

【0017】

The digital data conversion function unit 3 monitors the manufacturing condition data and product performance data in the manufacturing process 1 and stores these data in the manufacturing history database 4. Among them, in the case of forms, etc., it is converted into an electronic file (form digitization). ), Accumulate in the manufacturing history database 4, monitor and accumulate in the manufacturing history database 4 in the case of sensor signals, pattern recognition and accumulate in the manufacturing history database 4 in the case of images, etc. In such cases, it has a function to convert it into digital data by a method such as quantification by a specific algorithm (quantification of the amount of sensation) and store it in the manufacturing history database 4.

【0018】

The manufacturing condition data 5 and the workmanship data 6 are stored in the manufacturing history database 4. Of these, the manufacturing condition data 5 includes, for example, the product serial number (serial number), date, temperature, humidity, resist viscosity, resist coating strength, resist coating amount, and the like, as shown in FIG. This manufacturing condition data 5 also includes the flow rate of the process gas and its pressure.

【0019】

The workmanship data 6 indicates, for example, the quality of the product from the result of pattern recognition.For example, in the manufacture of a color CRT, it indicates how much white is displayed on the screen when white is displayed. If it is not white, the cause of the defect, for example, the resist temperature is an inhibitory factor, is recorded. Therefore, as shown in FIG. 2, for example, the workmanship data 6 stores a defect code that is an obstacle to the defect if the product has a defect.

【0020】

When the automatic analysis function unit 7 monitors the workmanship data 6 accumulated in the manufacturing history database 4 and detects a quality inhibition event (inhibition event identified from a defective code) on the production line, the production condition data 5 is also taken into consideration. It extracts quality-inhibiting factors. For example, when performance data 6 in the manufacturing history database 3 is constantly monitored by software and a quality-inhibiting event such as a sudden increase in defects or a change in yield tendency is detected, manufacturing is performed. The correlation with the manufacturing condition data 5 in the history database 3 is automatically analyzed by various data mining algorithms such as statistical methods, AI (artificial intelligence), and machine learning stored in advance, and quality inhibition factor candidates are obtained from this analysis result. Has a function to automatically extract data.

【0021】

The quality improvement history database 8 stores the quality improvement history data obtained by digitizing the discovery of known causes of defects or reduction in yield and improvement cases as digital data. For example, the quality inhibitory factors, the contents of the quality inhibitory factors, and the data thereof. The content of improvement for the quality inhibitory factor is stored in the form of, for example, if-then. For example, as shown in Fig. 3, it consists of the date, the content of quality inhibition (defective code), the factor of inhibition, the content of improvement, and the result of improvement. ..

【0022】

The validity confirmation function unit (improvement history collation) 9 collates the workmanship data 6 (extraction result) extracted by the automatic analysis function unit 7 with the quality improvement history data stored in advance in the quality improvement history database 8. It has a function to confirm the validity.

【0023】

The validity verification function unit (simulation verification) 10 executes a simulation of the phenomenon related to the manufacturing line for the quality inhibitory factors extracted by the automatic analysis function unit 7, that is, a simulation for analyzing the basic physicochemical phenomenon related to the manufacturing of the target product. It has a function to verify the validity.

【0024】

The automatic feedback function unit 11 should change the manufacturing conditions of the manufacturing process 1 so as to improve the quality-inhibiting factor when the validity or the validity verification function unit 10 has the validity or the validity by the validity confirmation function unit 9. It has a function of feeding back the instruction to the manufacturing condition control unit 2. In addition, the instruction for changing the manufacturing conditions in the manufacturing process 1 may be utilized in the product design department 12.

【0025】

The feedback control unit 13 receives, for example, a local control target in the production line 1, for example, in response to a quality inhibition event in the production line detected by the workmanship data 6 stored in the production history database 4 or the automatic analysis function unit 7. It has a function to control only the resist temperature to an appropriate value and notify an alarm or the like depending on the content of the quality inhibition event.

【0026】

The automatic quality control device monitors the manufacturing condition data 5 and the product workmanship data 6 in the manufacturing line 1 and stores them in the manufacturing history database 4, and of these, the workmanship data 6 is monitored in the manufacturing line 1. When a quality inhibition event is detected, the quality inhibition factor is extracted by taking into account the manufacturing condition data 5, and the extraction result is compared with the quality improvement history data stored in advance to confirm its validity, and further, the quality inhibition factor is confirmed. To verify the validity by executing a simulation of the phenomenon related to the production line 1, and if there is such validity and validity, change the manufacturing conditions so as to improve the quality inhibitory factor and feed it back to the production line 1. A memory (storage medium) 14 for storing a program is provided.

【0027】

Next, the operation of the device configured as described above will be described.

【0028】

In the production line 1, for example, in the case of semiconductor production, the production conditions, such as temperature, humidity, resist viscosity, resist coating strength, resist coating amount, process gas flow rate, and its pressure are controlled to control the semiconductor. Is manufactured.

【0029】

During the operation of the manufacturing line 1, the digital data conversion function unit 3 monitors the manufacturing condition data and the product performance data in the manufacturing process 1 and stores these data in the manufacturing history database 4. In this case, in the case of forms, etc., they are converted into electronic files (forms are digitized) and stored in the manufacturing history database 4, in the case of sensor signals, etc., they are monitored and stored in the manufacturing history database 4, and in the case of images, etc., pattern recognition. Then, it is stored in the manufacturing history database 4, and in the case of human sensory amount, it is converted into digital data by a method such as digitization by a specific algorithm (quantification of the sensory amount) and stored in the manufacturing history database 3. At this time, the workmanship data 6 indicates the quality of the product from, for example, the result of pattern recognition, and if there is a defect in the product, a defect code that hinders the defect is stored.

【0030】

Next, the automatic analysis function unit 7 constantly monitors the workmanship data 6 accumulated in the manufacturing history database 4 by software, and quality inhibition events (inhibition events that can be identified from the defective code) on the manufacturing line, for example, suddenly. When a quality inhibition event such as an increase in defects or a change in yield tendency is detected, the correlation with the manufacturing condition data 5 in the manufacturing history database 4 is determined by, for example, a statistical method stored in advance, AI (artificial intelligence), machine learning, etc. Automatic analysis is performed by various data mining algorithms, and quality inhibitory factor candidates are automatically extracted from the analysis results.

【0031】

Next, the validation function unit (improvement history collation) 9 collates the workmanship data 6 (extraction result) extracted by the automatic analysis function unit 7 with the quality improvement history data stored in advance in the quality improvement history database 8. And confirm its validity.

【0032】

Next, the validity verification function unit (simulation verification) 10 simulates the phenomenon related to the manufacturing line for the quality inhibitory factors extracted by the automatic analysis function unit 7, that is, analyzes the basic physicochemical phenomenon related to the manufacturing of the target product. Run a simulation to verify its validity.

【0033】

Next, the automatic feedback function unit 11 sets the manufacturing conditions of the manufacturing process 1 so as to improve the quality-inhibiting factor when the validity or the validity verification function unit 10 has the validity by the validity confirmation function unit 9 or the validity is justified by the validity verification function unit 10. The instruction is fed back to the manufacturing condition control unit 2 to change.

【0034】

The feedback control unit 13 receives, for example, a local control target in the production line 1 in response to a quality inhibition event in the production line detected by the workmanship data 6 stored in the production history database 3 and the automatic analysis function unit 7. For example, only the resist temperature is controlled to an appropriate value, and depending on the content of the quality inhibition event, an alarm or the like is notified.

【0035】

As described above, in the first embodiment described above, the production condition data 5 and the workmanship data 6 in the production line 1 are monitored and stored in the production history database 4, of which the workmanship data 6 is monitored and the production line 1 is used. When a quality inhibition event is detected, the quality inhibition factor is extracted by taking into account the manufacturing condition data 5, and the extraction result is compared with the quality improvement history data stored in advance to confirm its validity, and further, the quality inhibition is further performed. For the factors, run a simulation of the phenomenon related to production line 1 to verify the validity, and if there is such validity and validity, change the manufacturing conditions so as to improve the quality inhibitory factor and feed it back to production line 1. Therefore, a series of cycles such as "monitoring and database of manufacturing history information-> monitoring quality information-> extracting quality-inhibiting factors-> confirming the validity / correctness of factors-> improving by feedback to manufacturing conditions" is softened. By continuous execution, it becomes possible to perform automatic quality control that effectively utilizes a large amount of data that far exceeds the data processing capacity of human systems and overcomes ambiguity that relies on empirical rules and intuition. It should be noted that the intermediate result of the automatic cycle by the above-mentioned software can also be used for local device feedback control, human system management, and the like. In addition, the verification result by simulation can be fed back to the next product design.

【0036】

(2) Next, a second embodiment of the present invention will be described with reference to the drawings. The same parts as those in FIG. 1 are designated by the same reference numerals, and detailed description thereof will be omitted.

【0037】

FIG. 4 is a block configuration diagram of an automatic quality control device on a manufacturing line. In the warning judgment function unit 20, the validity confirmation by the validity confirmation function unit 9 or the validity confirmation by the validity verification function unit 10 is collated with the quality improvement history data stored in the quality improvement history database 8 and the validity verification is performed. If it cannot be confirmed by the simulation in the functional unit 10, the function for presenting the warning and the quality inhibitory factor candidate and asking whether to change the manufacturing condition of the production line 1 in order to improve the quality inhibitory factor. have.

【0038】

With such a configuration, the manufacturing condition data 5 and the workmanship data 6 in the manufacturing line 1 are monitored and accumulated in the manufacturing history database 4, of which the workmanship data 6 is monitored to detect quality inhibition events in the manufacturing line 1. When it is detected, the quality-inhibiting factor is extracted by taking into account the manufacturing condition data 5, and the extraction result is compared with the quality improvement history data stored in advance to confirm its validity. After verifying the validity by executing the simulation of the phenomenon related to, if these validity and validity cannot be confirmed, the warning judgment function unit 20 presents the warning and the candidate for the quality inhibitory factor, and the quality is inhibited. Ask if you want to change the manufacturing conditions of production line 1 to improve the factors.

【0039】

As a result, when a human gives an instruction to change the manufacturing conditions of the production line 1 in order to improve the quality inhibiting factor, the manufacturing conditions are changed and fed back to the manufacturing line 1 so as to improve the quality inhibiting factor.

【0040】

Therefore, in the first embodiment, the quality inhibitory factor candidates automatically extracted by various data mining algorithms are difficult to understand from known concepts and cannot be inferred from general physicochemical phenomena, but they are truly quality inhibitory. If it is a factor, the improvement by automatic feedback will not be activated.

【0041】

On the other hand, according to the second embodiment, when the quality inhibitory factor candidate is automatically extracted, the warning and the quality inhibitory factor candidate in data processing are presented and the factor is solved. Since humans are asked to decide whether or not to change the manufacturing conditions, in addition to the effect of the first embodiment described above, the robustness of the system is improved by cooperating with the human system, and only the conventional human system is used. It is also possible to speed up the work of examining, discovering, and improving the quality-inhibiting factors that have been carried out.

【0042】

(3) Next, a third embodiment of the present invention will be described with reference to the drawings. The same parts as those in FIG. 4 are designated by the same reference numerals, and detailed description thereof will be omitted.

【0043】

FIG. 5 is a block configuration diagram of an automatic quality control device on a manufacturing line. The learning function unit 21 has a function of adding this improvement case to the quality improvement history data 8 when the manufacturing conditions are changed to improve the quality-inhibiting factor.

【0044】

With such a configuration, the manufacturing condition data 5 and the workmanship data 6 in the manufacturing line 1 are monitored and accumulated in the manufacturing history database 4, of which the workmanship data 6 is monitored to detect quality inhibition events in the manufacturing line 1. When it is detected, the quality-inhibiting factor is extracted by taking into account the manufacturing condition data 5, the extraction result is compared with the quality improvement history data stored in advance to confirm its validity, and the manufacturing line 1 is further examined for the quality-inhibiting factor. The validity of the phenomenon is verified by executing a simulation of the phenomenon related to the above, and if there is such validity and validity, the manufacturing conditions are changed so as to improve the quality-inhibiting factor and the data is fed back to the manufacturing line 1.

【0045】

If the validity and validity cannot be confirmed, the warning judgment function unit 20 presents the warning and the quality inhibitory factor candidate, and changes the manufacturing conditions of the production line 1 in order to improve the quality inhibitory factor. Ask if you want to. Then, when a human gives an instruction to change the manufacturing conditions of the production line 1 in order to improve the quality inhibiting factor, the manufacturing conditions are changed so as to improve the quality inhibiting factor and fed back to the manufacturing line 1.

【0046】

In this case, the learning function unit 21 adds this improvement case to the quality improvement history data 8 when the manufacturing conditions are changed to improve the quality-inhibiting factor.

【0047】

In this way, when the automatic quality control system and humans cooperate to improve a completely unknown quality inhibitory factor, if it is further accumulated as a factor discovery and improvement example of a known defect occurrence or yield reduction, the subsequent automatic quality The self-learning and performance improvement of the control system can be improved.

【0048】

Furthermore, since the self-learning and performance improvement functions of the quality improvement method can be realized, it is possible to detect and improve the quality impeding factors at an early stage from these results, and it is possible to construct a stable and highly reliable production line. In addition, there is a possibility of discovering quality-inhibiting factors that were previously hidden, and improvements in quality and yield can be expected.

【0049】

The present invention is not limited to the first to third embodiments described above, and can be variously modified at the implementation stage without departing from the gist thereof.

【0050】

Further, the above-described embodiment includes inventions at various stages, and various inventions can be extracted by an appropriate combination of a plurality of disclosed constituent requirements. For example, even if some constituent requirements are deleted from all the constituent requirements shown in the embodiment, the problem described in the column of the problem to be solved by the invention can be solved, and the problem described in the column of the effect of the invention is described. If the above effect is obtained, the configuration in which this configuration requirement is deleted can be extracted as an invention.

【0051】

For example, in the third embodiment described above, when the manufacturing conditions are changed to improve the quality-inhibiting factor, this improvement case is added to the quality improvement history data 8, but the present invention is not limited to this. Data on manufacturing conditions that should not be performed in manufacturing may be accumulated in the quality improvement history data 8. As a result, it is not necessary to select the manufacturing conditions that should not be performed, and a more stable and highly reliable manufacturing line can be constructed.

【0052】

[Effect of the invention]

As described in detail above, according to the present invention, an automatic quality control method and an apparatus for a production line that can effectively utilize a large amount of data by overcoming ambiguity that relies on human data processing ability, empirical rules, and intuition. A storage medium can be provided.

[Simple explanation of drawings]

[Figure 1]

The block block diagram which shows 1st Embodiment of the automatic quality control apparatus of the manufacturing line which concerns on this invention.

[Figure 2]

The schematic diagram which shows the manufacturing condition data and the workmanship data in the 1st Embodiment of the automatic quality control apparatus of the manufacturing line which concerns on this invention.

[Fig. 3]

The schematic diagram which shows the quality improvement history database in 1st Embodiment of the automatic quality control apparatus of the manufacturing line which concerns on this invention.

[Fig. 4]

The block block diagram which shows the 2nd Embodiment of the automatic quality control apparatus of the manufacturing line which concerns on this invention.

[Fig. 5]

The block block diagram which shows the 3rd Embodiment of the automatic quality control apparatus of the manufacturing line which concerns on this invention.

[Explanation of symbols]

1: Manufacturing process (manufacturing line) 2: Manufacturing condition control unit 3: Digital data conversion function unit 4: Manufacturing history database 7: Automatic analysis function 8: Quality improvement history database 9: Validity confirmation function (improvement history collation) 10: Validity verification function (simulation verification) 11: Automatic feedback function 12: Product design department 13: Feedback control unit 14: Memory 20: Warning judgment function unit 21: Learning Function Department