矽源、觸媒與化學前處理對SiCN薄膜沉積與性質之影響

Abstract

本文是繼本研究室過去於SiCN研究對外加矽源的影響，進一步探討同時外加矽源和觸媒，化學蝕刻前處理對SiCN薄膜合成及性質之影響。合成採用微波電漿化學氣相沉積法在矽晶基材上以CH4和N2為氣體源，鍍有Co及Fe的矽棒為固體源，合成三元Si-C-N薄膜。合成之薄膜利用下列各項分析技術評斷之，包括掃描式及穿透式電子顯微鏡(SEM 和TEM)、電子繞射(ED)、X-ray繞射(XRD)、電子能譜化學分析儀(ESCA)、能量散佈能譜儀(EDS)、奈米壓痕儀（NIP）與I-V量測等。 在同時外加矽源與觸媒之時機方面，結果顯示其佔重要角色。同時施加矽源及觸媒可以分為兩種條件，亦即分為：(1)沉積前和，(2)沉積中兩種。於條件(1)下所得之薄膜顯示有較多再成核(re-nucleation)現象、晶體結構接近T-Si3N4、含Si量較少(Si4.3C12.9N4.0) 、有Si(2p)-C鍵結的存在、較低的奈米硬度與較佳之電子場效特性(電場強度為20 V/mm時，電流強度可達9360 mA/cm2)。相對而言，條件(2)下所得之薄膜，顯示晶體結構較接近a-Si3N4，並於晶粒層下方多一非晶質層。兩種條件下所得薄膜結構與性質之差異可由下列事實理解：條件(1)可等同於沉積氮化碳(CN)於CoSiX基材上之狀態。由於矽源較少而使得C原子與N原子反應機率提高。條件(2)下提供較多的矽源而大量地稀釋了Co的觸媒效應，因此較多的C原子位置被Si原子所取代，導致於無法偵測到Si(2p)-C鍵結。此項結果與文獻報導一致：認為矽-碳之鍵結需要1100 ℃以上之溫度才能發生。對照之下，以Co為觸媒且為條件(1)之下，矽-碳鍵結明顯，顯示Co觸媒有降低能障之效應。條件(2)也給予矽基材與氣體較多的反應機會，形成一層非晶質膜。 在前述條件(2)之下，實驗結果顯示觸媒種類(Co或Fe)對於SiCN之晶體結構、成分與鍵結有很大的影響。Co和Fe兩種觸媒比較下，以Fe為觸媒所得之薄膜晶體結構接近a-Si3N4、具<110>優選取向(preferred orientation)、含Si較高、含C較低(Si1.7C0.8N4.0)。以Co為觸媒也顯示含C量較高(Si2.7C4.6N4.0)，電子場效發射特性較佳(電場強度為20 V/mm時，電流強度為4710 mA/cm2)，並且優於Fe觸媒及本研究室過去只外加矽源之條件。 就化學蝕刻前處理對薄膜成核密度之影響，結果顯示無化學蝕刻者成核密度最高，其次為15分鐘者，然後為30分鐘化學蝕刻前處理者。此結果可由文獻結論來理解：認為成核密度有一最佳基材平均粗糙度之存在。不同或相同的腐蝕溶液，對於試片晶面有不同之蝕刻速率，因此每一蝕刻液可能產生不同表面粗度。報導之最佳粗糙度約在Ra = 20 nm左右，這個數值遠低於本研究中之表面平均粗度值。依此推斷，最佳平均表面粗糙度尚未獲得，可能低於15分鐘蝕刻時間。

The thesis followed the previous SiCN research in this laboratory on effect of additional Si source. Effects of simultaneously adding additional Si source and catalyst, and chemical etching pretreatment on SiCN deposition and properties were studied. The MPCVD method was used to synthesize the ternary SiCN films on Si substrate with CH4 and N2 as gas sources, with Si columns coated with Co or Fe catalysts on one side as solid sources. The films were characterized by the following analysis techniques, including scanning and transmission electron microscopy (SEM and TEM), electron diffraction (ED), X-ray diffraction (XRD), electron spectroscopy for chemical analysis (ESCA), energy dispersive spectrometry (EDS), nano-indentation probe (NIP) and I-V measurement. On effects of timing of Si source and catalyst applications, the results show it plays an important role. The simultaneous applications of both Si source and catalyst can be divided into two conditions, i.e., either: (1) before or (2) during films deposition. The films deposited by condition (1) show crystals with more re-nucleation phenomena, closer to T-Si3N4 crystal structure, less Si content (Si4.3C12.9N4.0), existence of Si (2p)-C bonding, lower nano-hardness and better field emission properties (9360 mA/cm2 at 20 V/mm). In contrast, the films by condition (2) reveal the crystals with structure closer to a-Si3N4 and show an additional amorphous layer under the crystalline layer. The differences in structure and property between the films deposited by two conditions can be reasoned by the following facts. The condition (1) is close to the condition of carbon nitride deposition on CoSiX substrate, and gives more chance for C atoms to react with N atoms due to less Si source. The condition (2) supplies more Si source and greatly dilutes the catalytic effect of Co; therefore, more C sites are replaced by Si atoms, and results in no detectable Si (2p)-C bonding. This evidence is in agreement with the reported statement that Si-C bonding can merely formed at temperatures higher than 1100 ℃. In contrast, the Co catalyst under condition (1) results in obvious Si-C bonding in the film, implying the Co catalyst can reduce the energy barrier of Si-C formation. The condition (2) also gives more chance for the substrate to react with gases to form an addition amorphous layer. The difference in nano-hardness between two conditions may relate to difference in crystal structure. Under condition (2) of the previous paragraph, the experimental results show that the type of catalysts (Co and Fe) has a great influence on crystal structure, composition and bonding of the films. In comparison between Co and Fe catalysts, the Fe catalyst gives rise to a crystal structure of the films closer to a-Si3N4 and with <110> preferred orientation, higher Si and lower C contents (Si1.7C0.8N4.0). The Co catalyst also shows a higher C content (Si2.7C4.6N4.0), and better field emission property (4710 mA/cm2 at 20 V/mm) than that of Fe catalyst or merely using additional Si source in our previous report in this laboratory. On effects of chemical etching pretreatment on nucleation density, the results show that the highest nucleation density is for the film without chemical pretreatment, the next for the film with 15 mins and then 30 mins chemical pretreatments. The results can be reasoned by the reported conclusion of existence of optimum substrate average roughness to obtain the highest nucleation density. Etching rate of each crystallographic plane is different for different solutions or the same solution. Therefore, each etching solution may produce different surface roughness. The reported optimum average roughness is around Ra = 20 nm, which is much lower than the present values. According to this conclusion, the optimum roughness has not been achieved in this experiment. The best etching time may be below 15 min.