以脈衝直流電源濺鍍氮化鈦薄膜之品質分析及機台實務介面整合

本文主要以脈衝直流磁控濺鍍系統鍍製氮化鈦薄膜，並使用田口實驗法萃取實驗中最佳化氮化鈦薄膜參數，分析得出最佳化鍍率、粗糙度及硬度之結論，進而建立模糊資料庫，並利用圖控軟體LabVIEW建立人機介面並達到即時監控之目的。本實驗使用田口L9直交表，挑選功率、工作頻率、氬氣流量、氮氣流量四個控制因子並分別設定三個水準。並針對鍍膜速率及薄膜硬度進行望大分析，對薄膜粗糙度進行望小分析等品質項目，進行實驗分析推算製程最佳化參數；在得出了各個目標之最佳化參數後，為了考慮兼顧鍍膜速率、薄膜粗糙度及薄膜硬度，導入一半準則，經過整合可得知功率(700 W)、頻率(25 kHz)、氬氣(7 sccm)及氮氣(4 sccm)為本實驗之製程最佳參數組。在薄膜相鑑定及沉積之晶體排列方向的驗證，係使用低掠角XRD進行分析，可清楚得出在2θ = 61.8°有一處(220) TiN的繞射峰；針對所設計的製程參數，其鍍膜速率介於0.42 Å/s ~ 0.95 Å/s；薄膜硬度則係使用奈米壓痕儀測定，其薄膜硬度介於15 ~ 21 GPa；薄膜粗糙度係使原子力顯微鏡測定，其薄膜粗糙度並介於0.4 ~ 0.8 nm。在目前人機介面中已完成關於電源供應器的一切控制及膜厚計的參數設置。因田口最佳化參數通常不會出現在設計的L9直交表中，所以必須經由田口實驗法得出之結論，進而建立模糊資料庫，而氣體流量控制器尚未整合，因此在模糊規則庫的輸出部分只有考慮了功率因子及頻率因子；而在奈米薄膜硬度機及AFM沒辦法即時與電腦整合，因此未納入模糊規則庫中，未來將進行更多實驗，將更完善規則建構至模糊規則庫中，即可利用模糊規則庫運算出結果達到控制的目的。

關鍵字

脈衝直流；田口實驗法；參數最佳化；硬度；粗糙度；模糊控制器

並列摘要

This paper aims at exploring the pulsed DC magnetron sputtering system in disposition application of titanium nitride film. In addition, the Taguchi experiments have been used to extract the optimized parameters of titanium nitride film, and make an analysis of the optimal deposition rate, roughness and hardness in conclusions. Furthermore, a fuzzy database is established and graphic software LabVIEW is applied to build human-machine interface for real-time monitoring. In this experiment, Taguchi L9 orthogonal array, power selection, operating frequency, argon flow rate, nitrogen flow rate have been used as four control factors. Moreover, three levels are also respectively set. The-larger-the-better analyses of deposition rate and film hardness have been conducted. The smaller-the-better analyses of film roughness in the category of quality have been conducted so as to analyze and calculate the optimized parameters. After getting the optimized parameters for individual objectives, in consideration of deposition rate, film roughness and film hardness, half of the criteria can be introduced. In integration, power (700 W), frequency (25 kHz), argon (7 sccm) and nitrogen (4 sccm) can be obtained. It may serve the optimized parameters on the production process in this experiment. In the film phase identification and the validation of deposited crystalline arrangement and direction, a low grazing-angle XRD analysis used for analysis can clearly show at 2θ = 61.8°, there is a (220) TiN diffraction peak. In view of designed process parameters, the deposition rate ranged from 0.42 Å / s to 0.95 Å / s; films harness is measured by nano-indentation hardness, showing its film hardness is between 15 to 21 GPa. Film roughness is measured by atomic-force microscopy, and its film roughness is between 0.4 ~ 0.8 nm. At the aspect of power supply control and parameter settings of the film thickness meter, currently the man-machine interface has been completed. Since optimized parameter for Taguchi does not normally appear in the design of the L9 orthogonal array, the conclusion must therefore be drawn by the Taguchi method. A fuzzy database will be further established. Meanwhile, the gas flow controller has not integrated, consequently the output part of the fuzzy rule data, which is only in consideration of the power factor and the frequency factor. Nano-film hardness device and AFM are not capable of have a real-time integration with the computer so they not included in the fuzzy rule database. For the future, more experiments will be conducted to build up more comprehensive rules into the fuzzy rule database. By doing so, the fuzzy rule database can successfully calculate the results for a better control.