標題: 銅金屬薄膜之抗氧化保護膜與擴散障礙層膜在極大型積體電路之銅金屬化應用上的可靠性研究
Reliability Issues of Passivation Layers against Copper Oxidation and Barrier Layers against Copper Diffusion in Copper Metallization for ULSI Application
作者: 莊瑞璋
Jui-Chang Chuang
陳茂傑
Mao-Chieh Chen
電子研究所
關鍵字: 銅金屬薄膜;抗氧化保護膜;擴散障礙層膜;鉻金屬;鉭金屬;鉬金屬;抗銅氧化能力;抗銅擴散能力;Cu oxidation;passivation layer;diffusion barrier;Cr;Ta;Mo;passivation capability;barrier capability
公開日期: 1998
摘要: 本論文探討抗銅氧化之保護膜與抗銅擴散之障礙層膜。前者係以200埃(A)厚度的鉭或鉻及其氮化物覆蓋之銅膜在氮氣或氧氣中高溫熱處理,之後檢視是否有氧化銅存在,並以未發現氧化銅的最高熱處理溫度為保護膜的抗銅氧化能力。後者係以500埃(A)厚度的鉻或鉬及其氮化物作為(銅/障礙層/p+n)接面二極體之抗銅擴散障礙層,並在氮氣中高溫熱處理之後,檢視其反偏漏電流,並以未發現反偏漏電流增大的最高熱處理溫度為障礙層的抗銅擴散能力。 在抗銅氧化保護膜的研究方面,濺鍍鉭膜與反應式濺鍍氮化鉭膜在氮氣中均可承受30分鐘700℃的高溫熱處理而不致使所覆蓋的銅膜氧化;同時鉭與氮元素的原子濃度和化學組態也不因700℃的高溫熱處理而有所變化。在氮氣中的高溫熱處理促使晶粒成長並且修復因濺鍍而造成的薄膜損傷,但高溫熱應力也在鉭基薄膜上產生新的缺陷。在氧氣中,濺鍍鉭覆蓋的銅膜可承受50分鐘400℃的高溫熱處理而不致使銅膜氧化;使用適當條件反應式濺鍍的非結晶態氮化鉭則可以增進抗銅氧化能力。然而不適當的反應式濺鍍條件,如過量的氮氣會造成過量氮電漿,使濺鍍的氮化鉭出現薄膜缺陷,而降低抗銅氧化能力;此舉甚至會因保護膜厚降低而加劇。此外,只要銅被氧化,氧化銅必然只出現在保護膜的最外層,這代表氧化過程中,是銅而非氧,擴散穿過鉭基保護膜。 鉻基保護膜包含濺鍍Cr-O與反應式濺鍍Cr-N-O膜在氮氣中均可承受30分鐘700℃的高溫熱處理而不致使所覆蓋的銅膜氧化;鉻基保護膜在氮氣中的高溫熱處理本質上是一種氮化與回火處理;而且700℃的氮氣高溫熱處理也沒有引發鉻與銅間的交互作用。在氧氣中,反應式濺鍍Cr-N-O覆蓋的銅膜可承受50分鐘500℃的高溫熱處理而不致使銅膜氧化,比濺鍍Cr-O覆蓋的銅膜高約150℃。其間的差異主要來自程度不同的Cr-O與Cr-N-O既存的薄膜缺陷。 濺鍍鉭覆蓋的銅膜經氮氣高溫熱處理,或稱氮氣預熱處理後其抗銅氧化能力大幅衰退,可能是鉭晶粒因為氮氣預熱處理而變大,因而縮短了銅的擴散路徑。相對地,以23.5 at%氮含量的反應式濺鍍氮化鉭覆蓋的銅膜,作300℃的氮氣預熱處理可提高其抗銅氧化能力到450℃;而且因氮氣預熱處理修復非結晶態氮化鉭膜的先天缺陷,而高溫熱應力產生的薄膜缺陷也被氮粒子塞滿,所以在本研究的溫度範圍內,氮氣預熱處理溫度愈高,其抗銅氧化能力愈好。但是,欲提高以氮含量為30.5 at%的反應式濺鍍氮化鉭膜覆蓋的銅膜之抗銅氧化能力則需提高氮氣預熱處理溫度至500℃以上,因為這一種氮化鉭的先天缺陷比前者嚴重。 鉻基保護膜,包含濺鍍Cr-O與反應式濺鍍Cr-N-O,所覆蓋的銅膜經氮氣預熱處理後其抗銅氧化能力均大幅衰退,因為Cr-O膜較嚴重的先天缺陷無法單純以氮氣預熱處理修復;而Cr-N-O膜雖然可因含氮粒子之塞滿缺陷而提高其保護能力,但是氮氣預熱處理在Cr-N-O膜產生孔洞,抵消前者的效應。因此,對鉻基保護膜而言,氮氣預熱處理是一道應該避免的程序。 在抗銅擴散障礙層膜的研究方面,濺鍍鉻膜可承受30分鐘500℃熱處理而不致使(銅/障礙層/p+n)接面二極體之反偏漏電流增大。控制濺鍍混合氣體中氬氣/氮氣流量比於24/6到24/12範圍,反應式濺鍍氮化鉻膜可具更好的障礙能力;而以氬氣/氮氣流量比為24/9濺鍍的氮化鉻膜則可承受30分鐘700℃熱處理而不造成二極體電氣特性劣化。以氬氣/氮氣流量比控制在24/8到24/12範圍所得的反應式濺鍍氮化鉬膜則可承受30分鐘600℃的熱處理而不造成二極體電氣特性劣化,比濺鍍鉬膜高100℃。氮粒子塞滿晶界的效應應是氮化鉻與氮化鉬膜的障礙能力分別高於鉻與鉬膜的原因。然而,在濺鍍混合氣體中過大之氮氣流量所造成的過量高能氮電漿衝擊已濺鍍之氮化物,易於造成鍍膜表面龜裂,導致障礙能力衰退。障礙層的失效模式有二,分別是:晶界擴散模式和區域性缺陷(微裂縫)擴散模式。前者適用於濺鍍鉻與鉬膜和低氮氣流量反應式濺鍍之氮化鉻與氮化鉬膜;後者適用於高氮氣流量反應式濺鍍之氮化鉻與氮化鉬膜;採用適當氬氣/氮氣流量比的反應式濺鍍氮化鉻與氮化鉬膜之失效模式則兼具上述兩者,但卻具有最佳的抗銅擴散能力。 快速回火處理可有效地增進鉻膜抗銅擴散之障礙能力,最佳快速回火處理溫度範圍為400℃至600℃。使用400℃快速回火處理的鉻障礙膜製成之(銅/障礙層/p+n)接面二極體可承受30分鐘700℃後續熱處理而不造成二極體電氣特性劣化,比使用未經快速回火處理的鉻膜所製成之二極體高出了200℃。在快速回火處理過程中,氮氣進入並填塞鉻之晶界,因而增進鉻膜的障礙能力。對鉬膜而言,700℃以下的快速回火處理未能有效增進鉬膜的抗銅擴散能力,似乎只能稍微改善或修復濺鍍鉬金屬的既存缺陷;但是800℃的快速回火處理卻可將鉬氮化成多晶的MoN型態,而且使用800℃快速回火處理的鉬障礙膜之二極體可以承受750℃的後續熱處理而不造成電氣特性劣化。800℃快速回火處理所生成的多晶MoN障礙膜應是抗銅擴散障礙能力增進的主因。
This thesis studies the passivation capability against Cu oxidation for the Ta-based and Cr-based passivation layers of 200A thickness as well as the barrier capability against Cu diffusion for the Cr-based and Mo-based barrier layers of 500A thickness. The passivation layers were sputter-deposited or reactively sputter-deposited on the Cu/SiO2/Si substrates, while the barrier layers were sputter-deposited or reactively sputter-deposited to make a structure of Cu/barrier/p+n junction diodes. Electrical measurements and various material analyses were used to investigate the passivation capability and barrier capability of these layers. Sputter-deposited and reactive sputter-deposited passivation layers were studied with respect to the passivation capability against oxidation of Cu in nitrogen and/or oxygen ambients at various annealing temperatures. For the Ta-based passivation layers, thermal annealing in N2 ambient at temperatures up to 700℃for 30min did not result in detectable Cu oxidation; and the atomic concentrations as well as the chemical states of Ta and N in the Ta and Ta-nitride films were not changed, either. The thermal N2 annealing resulted in the grain growth and the healing of the sputter-induced damage in the films, but it also induced new defects in the Ta-nitride films due to thermal stress. Thermal annealing in O2 ambient at temperatures up to 400℃for 50 min did not result in oxidation of the Ta-passivated Cu. This can be further improved by using a passivation layer of Ta-nitride film sputter-deposited in an appropriate condition; amorphism of the Ta-nitride film was presumably responsible for this improvement. However, sputter-induced surface damage by excess N2 plasma in the sputtering gas mixture may degrade the passivation capability of Ta-nitride films, and this effect can be enhanced in the case of very thin films. When the Ta or Ta-nitride passivated Cu was oxidized, the Cu oxides (CuO or Cu2O) always existed on the outermost surface. This suggests that it was copper, not oxygen, that diffused through the passivation layer during thermal O2 annealing of the Ta passivated Cu films as well as the Ta-nitride passivated Cu films. For the Cr-based passivation layers, both Cr-O and Cr-N-O passivation layers were able to prohibit oxidation of Cu at temperatures up to 700℃for 30min in N2 ambient. The thermal N2 treatment is essentially a nitrification and sintering process; even after N2 annealing at 700℃, there was presumably negligible interaction between Cr and Cu. In O2 ambient, the passivation capability of Cr-N-O layer was found to be 500℃, which is 150℃ higher than that of Cr-O layer. The distinction of the passivation capability against Cu oxidation was presumably due to the inherent defects in the Cr-O film, such as cracks and voids, which provide fast paths for oxygen and copper diffusion, and the nitrogen doping in the Cr-N-O film which decorated the surface defects and grain boundaries. The effects of thermal N2 annealing, or pre-sintering, on the passivation capability of Ta-based and Cr-based layers against Cu oxidation in O2 ambient was also investigated. For the Ta-based passivation layers, Ta layers with N2 pre-sintering at temperatures of 300℃and above revealed degradation in passivation capability, presumably because thermal N2 pre-sintering resulted in grain growth for the Ta passivation layer, leading to shorter paths of diffusion for Cu along the grain boundaries. In contrast, the nitrogen doped Ta-nitride layers showed a contrary trend. For the Ta-nitride layer with 23.5 at% (atomic concentration) of nitrogen, passivation capability was effectively improved by N2 pre-sintering at 300℃; the N2 pre-sintered sample was able to sustain thermal annealing in O2 ambient at temperatures up to 450℃without Cu oxidation. The healing of the sputter-induced damage in the Ta-nitride films by the thermal N2 pre-sintering was presumably responsible for the improvement of the passivation capability. Moreover, the higher the N2 pre-sintering temperature was, the better the result of passivation became. For the Ta-nitride layer with 30.5 at% of nitrogen, N2 pre-sintering at higher temperatures (500 to 700℃) was necessary to improve the passivation capability because the as-deposited Ta-nitride layer was more seriously damaged. For the Cr-based passivation layers, the passivation capability against Cu oxidation in O2 ambient was degraded after the N2 pre-sintering treatments. The higher the N2 pre-sintering temperature was, the lower the oxidation temperature of Cu became. The inherent defects in the Cr-O layers were not healed by N2 pre-sintering; instead, their sizes might grow with the increase of N2 pre-sintering temperatures. The unhealed defects of N2 pre-sintered Cr-O layers were presumed to be the principal reason of degradation for the Cr-O layers. On the other hand, voids formation after N2 pre-sintering at elevated temperatures was regarded as the cause of degradation for the Cr-N-O layers. Nitrogen in the Cr-N-O layers decorated the grain boundaries of Cr-nitride and improved the surface morphology of the layers, resulting in a better resistance against Cu oxidation than the Cr-O layer. However, the beneficial effect of nitrogen doping was outweighed by the formation of voids during the N2 pre-sintering process. Thus, the N2 pre-sintering treatment is an excess thermal treatment both for the Cr-O and Cr-N-O films, and it should be avoided for the application of Cr-O or Cr-N-O film as passivation layer against Cu oxidation. The barrier capability of sputter-deposited and reactively sputter-deposited Cr-based and Mo-based films against Cu diffusion in a structure of Cu/barrier/p+n junction diodes was studied. For a 500A thick Cr-based barrier layer, the barrier capability of a pure Cr layer was found to be 500℃, while Cr-nitride films sputter-deposited in a gas mixture of Ar and N2 showed improved barrier capabilities. With Ar/N2 flow rates of 24/6 to 24/12 sccm, the deposited Cr-nitride films possessed a much improved barrier capability. In particular, the Cu/Cr-nitride(24/9)/p+n junction diodes were capable of sustaining a 30 min of thermal anneal at temperatures up to 700℃without degradation of the diodes electrical characteristics. For the Mo-based barrier layer, the Mo-nitride films sputter-deposited in a gas mixture of Ar and N2 with Ar/N2 flow rates of 24/8 to 24/12 sccm were found to possess the best barrier capability. With a 500A thick Mo-nitride barrier layer, the Cu/Mo-nitride/p+n junction diodes were capable of withstanding 30 min of thermal annealing at temperatures up to 600℃without causing degradation to the devices electrical characteristics, which is a 100℃improvement over that of a 500A thick Mo barrier layer. The nitrogen decoration in the grain boundaries of Cr-nitride as well as Mo-nitride layers was presumably responsible for the improvement. However, for the Cr-nitride and Mo-nitride films deposited in a gas mixture of high nitrogen content, sputter-induced cracks formed in the Cr-nitride as well as Mo-nitride layers. At temperatures exceeding the thermal stability limit, the failure of Cu/barrier/p+n junction diodes was presumed to arise from two mechanisms: grain boundary diffusion for the lightly nitrogen doped Cr- and Mo-nitride as well as pure Cr and Mo barriers, and localized defect (micro-crack) diffusion for the excessively nitrogen doped Cr-nitride and Mo-nitride barriers. For the nitride films deposited in a gas mixture of medium nitrogen content (e.g., N2 flow rate of 6 to 12 sccm for Cr-nitride and 8 to 12 sccm for Mo-nitride), both of the failure mechanisms might exist simultaneously, but the films turned out to be the best with respect to the barrier capability against Cu diffusion. Rapid thermal annealing (RTA) of Cr films in NH3 ambient was found to be effective in improving barrier capability of Cr films in Cu metallization system. The improvement of barrier capability for Cr layers is closely related to the RTA temperature, and the optimum temperature was determined to be 400 to 600℃. The Cu/Cr/p+n junction diodes whose Cr barrier was RTA treated at a temperature of 400℃were able to sustain a 30min thermal annealing at temperatures up to 700℃without causing degradation to the diodes’ electrical characteristics, which is a 200℃improvement over the junction diodes using a Cr barrier without RTA treatment. It is believed that the nitrogen decoration in the grain boundaries of Cr layer was responsible for the improvement of the barrier capability. The RTA treatments of Mo layers at and below 700℃did not result in valuable improvement on the thermal stability of Cu/Mo/p+n junction diodes; presumably, these RTA treatments only resulted in healing the sputter-induced defect in the Mo films. However, the Cu/barrier/p+n junction diodes using a 800℃-RTA-treated Mo barrier layer were capable of withstanding a 30 min thermal annealing at temperatures up to 750℃without causing degradation to the devices electrical characteristics, which is a 250℃improvement over the Mo barrier layer without RTA treatment. The 800℃RTA treatment resulted in the formation of a polycrystalline MoN phase decorated with a significant amount of nitrogen at the grain boundaries as well as having a high re-crystallization temperature. This is regarded as the principal cause of the barrier capability improvement.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT870428008
http://hdl.handle.net/11536/64288
Appears in Collections:Thesis