碳化矽基介電層與銅金屬整合之電性和阻障特性

Abstract

本論文探討多種碳化矽基（SiC-based）低介電常數介電阻障膜之基本特性以及使用這些介電膜作為銅帽阻障層和蝕刻停止層的銅梳狀電容（銅大馬士革結構）之電性表現。基本特性的探討包括各種三甲基矽烷基（3MS-based）非晶系SiCN（介電常數小於5.5）介電層、三甲基矽烷和四甲基矽烷基（3MS- and 4MS-based）非晶系SiC（介電常數小於5）介電層、三甲基矽烷基（3MS-based）非晶系SiCO（介電常數小於4.5）介電層，以及八甲基環狀四矽氧烷基（OMCTS-based）非晶系SiCO介電層之熱穩定性、物性和阻障特性。在銅大馬士革結構的電性表現方面，本論文針對以非晶系SiCN、非晶系SiC或非晶系SiCO單層介電膜以及非晶系SiCN/SiC雙層介電膜作為銅帽阻障層的銅梳狀電容之時間相關介電崩潰（TDDB）壽命以及漏電和崩潰機制加以探討。 首先，三甲基矽烷基（3MS-based）非晶系SiCN膜的介電常數隨碳和氫含量的增加與氮含量的減少而從5.4減小至3.5，而其熱穩定溫度（測試條件為氮氣環境退火30分鐘）也從550降低至500oC。逐減的介電常數可歸因於碳原子具有較弱的雙極性和離子性偏極化以及氫原子具有較弱的電子性偏極化，而逐減的熱穩定性則歸因於在高溫時有大量的碳氫分子團揮發。含有較大量碳與氫的非晶系SiCN膜展現較差的阻障特性和介電強度，此乃由於碳原子導致介電膜具有較為薄弱交叉結合的分子結構和孔洞度的增加，以及氫原子導致介電膜內含有大量氫相關缺陷，諸如Si-H+-Si氫橋和Si-H弱鍵等。再者，鋁/非晶系SiCN/銅（MIM）電容結構的銅表面在沈積三甲基矽烷基（3MS-based）非晶系SiCN膜之前，如果先經過氧氣或氮氣電漿處理，則該MIM電容會呈顯較大的漏電流和較低的崩潰電場，這是因為銅表面有不穩定態Cu-O氧化物和Cu-N氮化物的生成。 其次，四甲基矽烷基（4MS-based）非晶系SiC膜（介電常數為4）比三甲基矽烷基（3MS-based）非晶系SiC膜（介電常數為4.7）具有較高的碳含量，因而具有較低的介電常數和熱穩定溫度（500oC），而且這種較為多孔性的四甲基矽烷基（4MS-based）非晶系SiC膜的阻障特性和抗水能力也較差。利用三甲基矽烷基（3MS-based）非晶系SiCN（介電常數為5）與非晶系SiC（介電常數為4）的雙層介電膜（α-SiCN/α-SiC）取代三甲基矽烷基（3MS-based）非晶系SiCN（介電常數為5）單層膜作為銅梳狀電容的銅帽阻障層，可大幅改善銅梳狀電容結構的時間相關介電崩潰（TDDB）壽命；這是因為非晶系SiC的漏電流較小、與其接觸的銅表面不會有氮化現象，而且非晶系SiC對銅膜和金屬線間的有機矽玻璃介電層（OSG IMD）都具有較佳的附著力。銅梳狀電容結構的漏電流與非晶系SiCN/SiC雙層膜的厚度比率相關。以SiCN(50奈米)/SiC(2奈米)雙層膜作為銅帽阻障層的銅梳狀電容結構，比起具有較厚非晶系SiC的雙層膜電容結構，諸如SiCN(40奈米)/SiC(10奈米)或SiCN(30奈米)/SiC(20奈米)，展現遠為優異的極低漏電流；較厚SiC膜電容結構的較大漏電流（屬於Frenkel-Poole激發機制）是因為較厚的非晶系SiC具有較大拉張力而易於產生較多界面缺陷，諸如在非晶系SiC/有機矽玻璃（OSG）界面的裂痕、孔洞、陷阱或斷鍵等。另一方面，銅梳狀電容結構的崩潰電場和時間相關介電崩潰（TDDB）壽命與非晶系SiCN/SiC雙層膜的厚度比率沒有明顯的關連性，而所觀察到的銅梳狀電容結構的崩潰是由於有機矽玻璃（OSG）本體發生介電崩潰。 第三，三甲基矽烷基（3MS-based）非晶系SiCO介電膜的介電常數隨著介電膜沈積製程中二氧化碳流量的增加（0至1200 sccm）而從4.4逐漸降至3.7；變小的介電常數可歸因於較大的氧含量導致介電膜具有較弱的電子性偏極化和較高的電負度。二氧化碳流量的增加也改善非晶系SiCO介電膜的熱穩定性、崩潰電場、漏電流、以及抗銅擴散能力；這是由於高流量二氧化碳沈積的三甲基矽烷基（3MS-based）非晶系SiCO介電膜具有較緻密和較少孔洞的結構。再者，以三甲基矽烷基（3MS-based）非晶系SiCO（介電常數為3.7）介電阻障膜作為介電層的銅金屬介電半導體（Cu-MIS）電容和作為銅帽阻障層的銅梳狀電容結構，都比用三甲基矽烷基（3MS-based）非晶系SiC（介電常數為4.4）介電膜製作的電容結構，展現大幅降低的漏電流和大幅提升的崩潰電場；此乃由於三甲基矽烷基（3MS-based）非晶系SiCO介電膜的結構密度較高、氧原子導致的薄膜特性改善、非半導體特徵、以及較低的邊緣或表面電場。八甲基環狀四矽氧烷基（OMCTS-based）非晶系SiCO膜的介電常數隨著介電膜沈積製程中的氧流量增加（0至300 sccm）而從2.8逐漸增至6.3；變大的介電常數可歸因於過量的氧原子導致介電膜具有較強的雙極性和離子性偏極化。氧流量的增加也會導致八甲基環狀四矽氧烷基（OMCTS-based）非晶系SiCO介電膜的熱穩定性、抗水能力、以及銅阻障特性之劣化。值得注意者，三甲基矽烷基（3MS-based）非晶系SiCO介電阻障膜比八甲基環狀四矽氧烷基（OMCTS-based）非晶系SiCO膜展現較佳的銅阻障特性；此乃由於三甲基矽烷基（3MS-based）非晶系SiCO介電膜具有碳化矽基（SiC-based）的分子結構，而八甲基環狀四矽氧烷基（OMCTS-based）非晶系SiCO介電膜的分子結構卻是氧化矽基（SiO-based）之故。

This thesis study includes the basic properties of various silicon-carbide-based (SiC-based) low-k dielectric barrier films as well as the electrical performance of Cu-comb capacitors (Cu damascene structure) using these dielectric films as the Cu cap-barrier and etching stop layer. The study of the basic properties includes the thermal stability and physical and barrier properties of various 3MS-based α-SiCN (k<5.5) dielectrics, 3MS- and 4MS-based α-SiC (k<5) dielectrics, 3MS-based α-SiCO (k<4.5) dielectrics, and OMCTS-based α-SiCO dielectric films. With regard to the electrical performance of the Cu damascene structure, the TDDB lifetime and the leakage and breakdown mechanisms are investigated using Cu-comb capacitors with a dielectric film of α-SiCN, α-SiC, or α-SiCO as well as with a bilayer dielectric stack of α-SiCN/α-SiC as a Cu cap-barrier layer. First, the dielectric constant of the 3MS-based α-SiCN films decreases from 5.4 to 3.5 with increasing carbon and hydrogen contents and decreasing nitrogen content in the dielectric films, and the thermally stable temperature of the α-SiCN films (annealed in N2 ambient for 30 min) is also decreased from 550 to 500oC. The reduced dielectric constant is attributed to the lower dipolar and ionic polarizations of carbon atoms and the lower electronic polarization of hydrogen atoms, while the decreased thermal stability is due to massive outgassing of hydrocarbon groups at elevated temperatures. The α-SiCN films with abundant carbon and hydrogen contents exhibit degraded barrier property and dielectric strength. This can be attributed to the carbon-atom-induced poorly crosslinked molecular structure and porosity enrichment as well as the hydrogen-related defects, such as Si-H+-Si hydrogen bridges, and the numerous Si-H weak bonds arisen from the hydrogen atoms in the dielectric film. Moreover, the Al/α-SiCN/Cu MIM capacitors with the Cu-surface exposed to O2- or N2-plasma treatment prior to the deposition of the 3MS-based α-SiCN film exhibit higher leakage current and lower breakdown field. The increased leakage current and decreased breakdown field of the O2- and N2-plasma-treated samples are attributed, respectively, to the presence of metastable Cu-O oxide and Cu-N azide at the Cu-surface. Second, the 4MS-based α-SiC (k=4) film contains a higher carbon content and thus has a lower dielectric constant and thermal stability temperature (500oC) than the 3MS-based α-SiC (k=4.7) film. The more porous 4MS-based α-SiC film also exhibits degraded barrier property and moisture resistance. The TDDB lifetime of Cu-comb capacitor is greatly improved by using a 3MS-based α-SiCN(k=5)/α-SiC(k=4) bilayer dielectric stack instead of the 3MS-based α-SiCN(k=5) single film as a Cu cap-barrier. This improvement is attributed to the lower leakage current of α-SiC, absence of nitridation on Cu surface, and better adhesion of α-SiC to Cu and OSG IMD. The leakage current between Cu lines in the Cu-comb capacitor is dependent on the thickness ratio of the α-SiCN/α-SiC bilayer barrier. The Cu-comb capacitor with an α-SiCN(50 nm)/α-SiC(2 nm) bilayer barrier exhibits a much smaller leakage current than that with a bilayer barrier of thicker α-SiC film, such as α-SiCN(40 nm)/α-SiC(10 nm) or α-SiCN(30 nm)/α-SiC(20 nm). The increased leakage (Frenkel-Poole emission) between Cu lines is attributed to the large number of interfacial defects, such as cracks, voids, traps or dangling bonds at the α-SiC/OSG interface, which are generated by the larger tensile force of the thicker α-SiC film. On the other hand, the breakdown field and TDDB lifetime of the Cu-comb capacitor reveal little dependence on the thickness ratio of the α-SiCN/α-SiC bilayer barrier, and the observed breakdown of the Cu-comb capacitor is due to dielectric breakdown of the bulk OSG layer. Thirdly, the dielectric constant of the 3MS-based α-SiCO dielectric films decreases from 4.4 to 3.7 with increasing CO2 flow rate (0~1200 sccm) during the dielectric deposition process. The decreased dielectric constant is attributed to the lower electronic polarization and higher electronegativity of oxygen atoms in the α-SiCO dielectric film. Increasing CO2 flow rate also leads to an α-SiCO dielectric film of better thermal stability, higher breakdown field, lower leakage current, and superior resistance to Cu diffusion. The improved barrier property is attributed to the denser and less porous structure of the 3MS-based α-SiCO dielectric film deposited with a higher CO2 flow rate. Moreover, the Cu-MIS and Cu-comb capacitors with a 3MS-based α-SiCO (k=3.7) dielectric barrier film exhibit much smaller leakage current and higher breakdown field than those with a 3MS-based α-SiC (k=4.4) dielectric film. This is attributed to the higher density, oxygen-improved film’s property, non-semiconductor behavior, and lower fringe- or surface-electric field of the 3MS-based α-SiCO dielectric film. The dielectric constant of the OMCTS-based α-SiCO films increases from 2.8 to 6.3 with increasing O2 flow rate (0~300 sccm) during the dielectric deposition process, presumably due to the increasing content of the higher dipolar and ionic polarized oxygen atoms in the dielectric film. Increasing O2 flow rate also degrades the thermal stability, moisture resistance, and Cu barrier property of the OMCTS-based α-SiCO dielectric film. Notably, the 3MS-based α-SiCO dielectric barrier film is superior to the OMCTS-based α-SiCO film in the Cu barrier property. This is due to the fact that the 3MS-based α-SiCO films have a SiC-based molecular structure, while the molecular structure of the OMCTS-based α-SiCO films is SiO-based.