Kim, Dong Hwan
(Department of Electronics Engineering, Myongji University)
,
Choi, Jeong Eun
(Department of Electronics Engineering, Myongji University)
,
Ha, Tae Min
(Department of Electronics Engineering, Myongji University)
,
Hong, Sang Jeen
(Department of Electronics Engineering, Myongji University)
Virtual metrology, which is one of APC techniques, is a method to predict characteristics of manufactured films using machine learning with saving time and resources. As the photoresist is no longer a mask material for use in high aspect ratios as the CD is reduced, hard mask is introduced to solve ...
Virtual metrology, which is one of APC techniques, is a method to predict characteristics of manufactured films using machine learning with saving time and resources. As the photoresist is no longer a mask material for use in high aspect ratios as the CD is reduced, hard mask is introduced to solve such problems. Among many types of hard mask materials, amorphous carbon layer(ACL) is widely investigated due to its advantages of high etch selectivity than conventional photoresist, high optical transmittance, easy deposition process, and removability by oxygen plasma. In this study, VM using different machine learning algorithms is applied to predict the thickness of ACL and trained models are evaluated which model shows best prediction performance. ACL specimens are deposited by plasma enhanced chemical vapor deposition(PECVD) with four different process parameters(Pressure, RF power, $C_3H_6$ gas flow, $N_2$ gas flow). Gradient boosting regression(GBR) algorithm, random forest regression(RFR) algorithm, and neural network(NN) are selected for modeling. The model using gradient boosting algorithm shows most proper performance with higher R-squared value. A model for predicting the thickness of the ACL film within the abovementioned conditions has been successfully constructed.
Virtual metrology, which is one of APC techniques, is a method to predict characteristics of manufactured films using machine learning with saving time and resources. As the photoresist is no longer a mask material for use in high aspect ratios as the CD is reduced, hard mask is introduced to solve such problems. Among many types of hard mask materials, amorphous carbon layer(ACL) is widely investigated due to its advantages of high etch selectivity than conventional photoresist, high optical transmittance, easy deposition process, and removability by oxygen plasma. In this study, VM using different machine learning algorithms is applied to predict the thickness of ACL and trained models are evaluated which model shows best prediction performance. ACL specimens are deposited by plasma enhanced chemical vapor deposition(PECVD) with four different process parameters(Pressure, RF power, $C_3H_6$ gas flow, $N_2$ gas flow). Gradient boosting regression(GBR) algorithm, random forest regression(RFR) algorithm, and neural network(NN) are selected for modeling. The model using gradient boosting algorithm shows most proper performance with higher R-squared value. A model for predicting the thickness of the ACL film within the abovementioned conditions has been successfully constructed.
* AI 자동 식별 결과로 적합하지 않은 문장이 있을 수 있으니, 이용에 유의하시기 바랍니다.
제안 방법
In this study, virtual metrology using different machine learning algorithms(gradient boosting, random forest, neural network) is applied to predict the deposition rate of ACL through plasma enhanced chemical vapor deposition(PECVD) and learned model are analyzed which model is most proper about ACL thickness data.
Although there are several methods to evaluate the modeling, it is wrong to evaluate the modeling with only one method in terms of its reliability. So, in this study, trained models are compared with three of different analysis methods which model is most proper about ACL data in given parameter ranges. To evaluate each model, root mean square error(RMSE), mean absolute percentage error(MAPE) and R-squared, which measure of prediction accuracy of a forecasting method, are used.
이론/모형
Also, it may contain meaningless data. So, to plan efficient process parameter value before fabrication and acquire valuable ACL film data, Box-Behnken method of design of experiment(DOE) utilize. Box-Behnken refers to analyze the data statistically and plan to obtain the maximum information with the minimum number of experiments [12].
However, ensemble tree method has advantages of handling missing value, obtaining finer-grain and generalized prediction model [11]. Thus, modeling is carried out using neural network and ensemble tree(random forest algorithm and gradient boosting algorithm).
So, in this study, trained models are compared with three of different analysis methods which model is most proper about ACL data in given parameter ranges. To evaluate each model, root mean square error(RMSE), mean absolute percentage error(MAPE) and R-squared, which measure of prediction accuracy of a forecasting method, are used.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.