本研究比較了在相同製程條件下，使用350 kHz脈衝直流和13.56 MHz射頻反應濺鍍技術在矽基板上沉積的氮化鋁(AlN)薄膜的質量，並針對不同電源進行改變。我們的重點是評估薄膜屬性如晶體性、表面形態和厚度均勻性的差異。我們的結果表明，雖然SEM橫斷面圖像顯示射頻濺鍍的沉積速率較低，但產生的薄膜展示出更好的結晶性，這一點從XRD峰的更尖銳和更強烈，特別是<002>方向的峰可以看出。從針對c軸取向AlN薄膜的XRD光譜導出的計算顯示，利用射頻電源，最佳的c軸晶格常數為4.8706 Å，而<002>方向的紋理係數(TC002)為1.99。這些值都超過了在相同功率下，脈衝直流電源得到的值。此外，原子力顯微鏡(AFM)的測量顯示，無論是射頻還是脈衝直流濺鍍沉積的薄膜，其表面粗糙度隨著濺鍍功率的增加而略有增加，均方根(RMS)值範圍在2.08 nm至5.98 nm之間。光發射光譜(OES)被用來研究在氮氣氛圍中被脈衝直流和射頻電源所激發電漿中的活性物質的生成。射頻濺鍍中觀察到的更高光譜強度，特別是在約390.93 nm處的氮離子(N2+)的特徵發射，種種研究都顯示射頻電漿提供了有利於形成高質量AlN薄膜的能量條件。;This study compares the quality of aluminum nitride (AlN) thin films deposited on Si substrates using 350 kHz pulsed DC and 13.56 MHz RF reactive sputtering techniques under identical process conditions with varied source power. Our focus was on evaluating differences in film properties such as crystallinity, surface morphology, and thickness uniformity. Our results indicated that, although SEM cross-sectional images showed a lower deposition rate for RF sputtering the produced films exhibit better crystallinity, as evidenced by sharper and more intense XRD peaks, particularly the <002> peak. Calculations derived from XRD spectra of c-axis oriented AlN thin films, utilizing RF source power, revealed that the optimal c-axis lattice constant is 4.9525 angstroms, and the texture coefficient in the <002> (TC002) direction is 1.99. Both values surpass those obtained at the same power level in pulse DC power source. Additionally, AFM measurements indicate that the surface roughness of films deposited by both RF and pulsed DC sputtering slightly increased with the sputtering power, with the root mean square (RMS) values ranging between 2.08 nm and 5.98 nm. Optical emission spectroscopy (OES) has been used to study the generation of reactive species in plasma excited by both pulsed DC and RF power sources in a nitrogen atmosphere. The higher spectral intensities observed in RF sputtering, particularly the emission characteristic of ionized nitrogen (N2+) at around 390.93 nm, suggest that RF plasma provides energy conditions conducive to the formation of high-quality AlN films. Additionally, we applied the PCA algorithm for big data analysis to reduce dimensionality and visualize clustering results of Optical Emission Spectroscopy (OES) data recorded during the deposition of thin films using two different power sources. It is evident from the analysis that there are distinct clustering effects for both power sources, thus substantiating the presence of diverse characteristics between the two sources.
