利用無電鍍金屬與矽化反應物為觸媒選擇性水平成長碳奈米結構之製程及其性質

Abstract

控制碳奈米結構的成長，仍為目前碳奈米技術主要的研究課題之一。本實驗主要目的為開發碳奈米結構製程並探討其性質，亦即研究選擇性水平成長碳奈米結構在具圖案的矽基材上。利用微波電漿化學氣相沉積法以CH4與H2為反應氣體沉積碳奈米結構。爲了研究不同觸媒沉積法對CNTs之影響，本研究主要分為兩種觸媒沉積法：一、無電鍍法析鍍在指定位置上(Ni、Co和Fe薄膜)；二、利用PVD法沉積在指定位置上並經RTA退火產生矽化反應物(Co和Ni矽化物)；目的在比較不同的觸媒不同的沉積方法之影響。以微影蝕刻之技術將矽晶片製成圖案，利用此圖案矽晶片可成功地水平成長碳奈米管並橫跨溝槽。將觸媒沉積在指定位置之選擇性係利用a:Si與Si3N4之間導電性的差異使無電鍍金屬在導體(a:Si)析出之特性與金屬和金屬只會與a:Si產生矽化物而不會與Si3N4反應之特性，使觸媒在指定位置(a:Si)反應析出。再經前處理(H電漿還原，H2 = 100 sccm，400 W，9 Torr，550 ~ 580℃)使觸媒形成奈米顆粒，接著利用MPCVD(CH4/H2 = 1 / 100 sccm/sccm，800 W，16 Torr，660 ~ 680℃，偏壓 = 0、-200 V)使CNTs選擇性沉積在觸媒指定位置上。此外圖案之Si3N4層是被設計來引導碳奈米管成長方向藉以水平成長橫跨溝槽兩側之間。實驗分析主要以ESCA、SEM、TEM、HRTEM、Raman、J-V 曲線，來依序分析碳奈米結構觸媒源之成份、表面形貌、微結構、晶格影像、鍵結以及場效發射性能。 結果顯示無電鍍Ni、Co、Fe觸媒成長出來的碳奈米管管長分別約為(100 ~ 1200 nm，200 ~ 1200 nm，200 ~ 1200 nm)，管徑分別約為 (10 ~ 30 nm，25 ~ 40 nm，20~50 nm)，管數密度分別為(1290，387，516 Gtubes/inch2)。以上結果顯示以Ni為觸媒成長碳奈米管有高的管數密度，但管長與管徑皆比較小。值得一提的是以無電鍍Ni、Co、Fe為觸媒成長之碳管微結構皆為基底成長(Base-growth)的竹節狀碳奈米管，表示了觸媒與基材之間鍵結較強，雖然一般而言以PVD製備之觸媒成長之碳奈米管為頂端成長(Tip-growth)；另外沒有添加N2在反應氣體中，卻形成竹節狀碳奈米管的原因可能是在無電鍍過程中，鍍液中強還原劑聯胺的N殘留在試片上所致。而基材負偏壓使無電鍍金屬觸媒成長之碳奈米管的管數密度明顯減少但管長與管徑輕微增加。結果顯示可經由操控無電鍍析鍍時間來改變碳奈米管之管數密度。 以PVD沉積法經RTA處理後之CoSix與NiSix觸媒成長碳奈米管，其管長分別約為(100 ~ 300 nm，300 ~ 1000 nm)，管徑分別約為(10 ~ 30 nm，30 ~ 40 nm)，管數密度分別為(13，6 Gtubes/inch2)。由Raman頻譜圖峰值ID/IG比(或缺陷密度)得知以矽化物為觸媒成長CNTs之缺陷密度比無電鍍金屬高(亦即CoSix> Fe > Co > Ni)。以場發射特性而言，不同觸媒成長碳奈米管之啟始電場依序如下CoSix> Co > Fe > Ni。 在圖案矽晶片Si3N4層對碳奈米管成長之引導效應方面，結果顯示若無引導層無法水平成長碳奈米管。而在圖案引導碳奈米管之成長效率方面，無電鍍金屬觸媒法明顯的高於矽化反應物法。原因可能在於無電鍍法屬於化學析鍍法，比起PVD法有方向性的沉積特性，鍍液較易深入析鍍在具圖案矽晶片之溝槽深處指定位置。

Controllable growth of carbon nanostructures is still one of main research topics of carbon nanotechnology at present. The main purpose of this study is to develop a fabrication process of the patterned carbon nanostructures and to examine their properties. The carbon nanostructures were deposited by microwave plasma chemical vapor deposition (MPCVD) method with CH4 and H2 as source gases. The catalysts for nanostructure growth were prepared by two different processes: (1) electroless plating process (Ni, Co and Fe thin films), and (2) physical vapor deposition (PVD) process and followed by silicide reaction in rapid thermal annealing (RTA) furnace (Co- and Ni-silicides). A process was successfully demonstrated to be able to grow the horizontally-oriented carbon nanotubes (CNTs) cross the trenches of the patterned Si wafer, which was produced by conventional photolithography technique. The selectivity of the process is based on the difference in electrical conductivity between amorphous silicon (a:Si) and Si3N4, where the catalyst by electroless process can only be deposited on a:Si substrate. The selectivity is also based on greater chemical reactivity of the catalyst with a:Si to form silicides, instead of with Si3N4. The catalyst pattern was then pretreated in H plasma (H2=100 sccm, 400 W, 9 Torr, and 550℃ ~ 580℃), and followed by CNTs deposition (CH4/H2 = 1 / 100 sccm/sccm, 800 W, 16 Torr, 660℃ ~ 680℃, and substrate bias = 0, -200 V) in MPCVD. The carbon nanostructures can selectively deposit on the areas of the pattern with metal catalyst or silicides. Furthermore, the Si3N4 layer of the pattern was so designed to guide the growth of carbon nanostructures in horizontal direction to bridge the trenches of the pattern. The as-deposited catalysts were examined by ESCA. The CNTs were characterized by SEM, TEM, HRTEM, Raman and field emission (J-V) measurements. For catalysts deposited by electroless process, the results show that the features of the CNTs for Ni, Co and Fe catalysts are (100 ~ 1200 nm, 200 ~ 1200 nm, 200 ~ 1200 nm) in length, (10 ~ 30 nm, 25 ~ 40 nm, 20 ~ 50 nm) in diameter, (1290, 387, 516) Gtubes/inch2 in tube number density, respectively. It indicates that the CNTs for Ni catalyst give a greater tube number density but smaller in length and diameter. It is interesting to note that the CNTs are bamboo-like and grown by base growth mode for Ni, Co and Fe catalysts, signifying a stronger bonding between catalyst and substrate, though the CNTs are generally the tip growth CNTs for catalysts deposited by PVD. It is also noted that the bamboo-like CNTs were formed without the introduction of N2 during CNTs deposition. The possible reason may be due to the residual nitrogen on specimen during electroless deposition in such a strong reducing agent. Effect of the substrate bias was found to cause a great decrease in the tube number density, but a slight increase in length and diameter. The results show that the tube number density can be manipulated by varying the deposition time of the plating process of the catalyst. For catalysts prepared by PVD and followed by RTA treatment, the features of the CNTs for CoSix and NiSix as catalysts are: (100 ~ 300, 300 ~ 1000 nm) in length, (10 ~ 30 nm, 30 ~ 40 nm) in diameter, (13, 6 Gtubes/inch2) in tube number density. In terms of ID/IG ratio (or defect density) of the Raman peaks for CNTs, the results show that CNTs with silicides as catalysts give rise to a higher defect density than pure metals (CoSix > Fe > Co > Ni). In terms of field emission properties, the turn-on voltages of the CNTs for different catalysts are in order of CoSix > Co > Fe > Ni. Regarding guiding effect of Si3N4 layer on CNTs growth, the results indicate that the horizontally-oriented CNTs could not be obtained without guiding. Furthermore, the guiding efficiency of the CNTs growth for catalysts deposited by electroless process is much better than that by silicide reaction. The reason may be due to a greater ability for a chemical solution in electroless process to penetrate into the trenches of the pattern, by comparing with PVD method, where the deposition is more directional.