高熵合金與其氮化物薄膜作為銅製程擴散阻障層之研究

On the effectiveness of high-entropy alloy/nitride thin films as diffusion barrier for copper metallization

蔡銘洪(Ming-Hung Tsai)

指導教授 ： 葉均蔚 ； 甘炯耀

國立清華大學/工學院/材料科學工程學系/博士(2008年)

摘要

本論文主題在探討高熵合金及其氮化物作為IC銅製程擴散阻障層的效果。我們針對擴散阻障的應用，設計一新的高熵合金成份AlMoNbSiTaTiVZr，先進行不同含氮量的(AlMoNbSiTaTiVZr)Nx薄膜之結構與機械性質的基礎了解。接著則分別使用AlMoNbSiTaTiVZr，(AlMoNbSiTaTiVZr)Nx及NbSiTaTiZr等三種薄膜進行擴散阻障的效果測試。最後，由於我們發現失效的擴散阻障測試試片表現經常會出現大量具特殊方向性的矽化物線段，然而文獻中卻未曾有人對此現象做探討，故而我們也研究了這些矽化物線段的生長機制。
不同含氮量(AlMoNbSiTaTiVZr)Nx薄膜部份，我們發現低氮氣通量時，(AlMoNbSiTaTiVZr)Nx薄膜內的氮含量隨氮氣通量上升而快速增加。但當氮氣通量(RN)到達33%及以上時，氮含量在則飽和在50%附近。結晶結構方面，低氮氣通量時(RN=0%和3%)，薄膜為非晶結構，此時其截面看起來也沒有特殊的形貌。氮氣通量為11%時，結構則為FCC奈米晶與非晶兩相共存。氮氣通量更高時，薄膜結構轉為FCC，並且其截面呈現出柱狀結構。機械性質方面，由於氮的加入會與其他金屬產生很強的鍵結，故而薄膜硬度隨氮含量增加而上升。AlMoNbSiTaTiVZr合金膜在所有沒有施加基板偏壓鍍膜的高熵合金膜中，具有最高的硬度。同樣的，其氮化物在所有沒有施加基板偏壓鍍膜的高熵合金氮化物膜中，也具有最高的硬度，這應與其具有較多元素而造成的效應有關。我們並認為，多元效應可以作為未來設計硬膜材料的一個可能方式。至於高熵合金/氮化物膜用作擴散阻障層的實際測試，我們將阻障層鍍在(100)矽基板與300 nm的銅之間，並在不同溫度進行30分鐘的退火。結果發現100 nm的AlMoNbSiTaTiVZr合金膜可耐至700 °C。70 nm的(AlMoNbSiTaTiVZr)50N50氮化物可耐至850 °C，是所有文獻中測試過的阻障層中表現第二好的，僅次於Ta-Si-N系列。20 nm的NbSiTaTiZr合金膜則可耐至800 °C。這些結果說明高熵合金/氮化物在這方面有相當優異的表現，有絕佳的化學穩定性，不會與矽反應產生矽化物，也有極好的抗擴散能力，能防止銅的擴散。它們的化學穩定是由於高熵合金/氮化物內的原子之間具有相當大的混合焓，而多種元素的混合則帶來大量的混合熵，而使得高熵合金具有較低的能量狀態，故較為穩定。至於它們能夠阻擋銅的擴散，是因為它們的非晶結構有絕佳的熱穩定性，比一般金屬/氮化物需要更高的溫度才會結晶化產生容易擴散的晶界。其非晶結構的熱穩定性來自於多元素的效應，多元素造成較大的晶格扭曲，而使得形成結晶結構的驅動力減少。多元素也會造成此非晶結構有較高的原子堆疊密度，使其內的擴散減緩，而使結晶更難發生。整體而言，高熵合金/氮化物的表現非常優異，可作為未來擴散阻障層材料設計的可行方向。
至於矽化物線段的生長機制方面，銅首先是由製程中髒污或是灰塵顆粒造成的阻障層缺陷點擴散進矽基板。藉著退火氣氛中氧的協助，使得銅原子變成能高速擴散的一價銅離子。由於銅在矽基板中的擴散異向性，沿著<110>和<110>方向，銅的濃度增加較快，故而矽化物會在這兩個方向先達到臨界濃度而優先孕核成長，因此形成沿著這兩個方向排列的矽化物顆粒線段。當矽化物逐漸長大，它們會膨脹凸出矽基板而推擠上方的阻障層和銅膜，導致阻障層變形並且裂開。阻障層一但裂開，銅即可在此處穿入，而沿著矽化物擴散進基板，故而等於增加了一個銅的來源。由於隨著線段成長，銅的供應點會不斷自動產生，因此線段的成長變得能夠自給自足而能不斷前進。由上述成長機制可知，矽化物長成具方向性的線段形貌，是因為銅的供應是來自於某些特定缺陷位置，再由這些位置往外擴散。這也就代表著，雖然試片表面有大量矽化物，但並非由於阻障層本身的失效而造成銅全面性的入侵，而是因為阻障層上的缺陷造成銅入侵，才往外擴散產生矽化物。所以，為了真正了解阻障層材料本身的熱穩定性，應特別注意阻障層的均勻性，以及退火氣氛中的氧氣含量，才能正確得到其材料的內稟失效溫度。

關鍵字

高熵合金 ； 擴散阻障層 ； 銅金屬化 ； 多元氮化物 ； 高熵材料 ； 硬膜

並列摘要

The aim of this study is to test the effectiveness of high-entropy alloy/nitrides as diffusion barrier layer for copper metallization. A new HEA composition, AlMoNbSiTaTiVZr, is specifically designed for this purpose. We first deposited various (AlMoNbSiTaTiVZr)Nx films with different nitrogen concentration, and investigated their structure and mechanical properties. Then three high-entropy alloy/nitride films including AlMoNbSiTaTiVZr, (AlMoNbSiTaTiVZr)50N50, and NbSiTaTiZr were tested to understand their effectiveness as diffusion barriers. In the characterization of (AlMoNbSiTaTiVZr)Nx, we found that the concentration of nitrogen increases rapidly at lower RN and then saturates near the stoichiometric composition at RN = 33% and above. At lower RN (0% and 3%), film structures are amorphous and their cross-sectional microstructures are featureless. At RN of 11%, FCC nitride nanocrystals coexist with amorphous phase. At higher RN ( ≧ 33% ), films exhibit single FCC NaCl-type structure having fine column structures with nanograins. Compared with other published HEA films prepared without bias, AlMoNbSiTaTiVZr exhibits the highest hardness and modulus both in its alloy and nitride states up to now. This confirms the effectiveness of the present alloy design. High-entropy alloy concept could provide another route to enhance mechanical properties in addition to the route of nanocomposite design. As for the effectiveness of high-entropy alloy/nitride as diffusion barrier, standard Cu/barrier/Si test structures were fabricated and annealed at different temperature for 30 min. The results show that a 100 nm AlMoNbSiTaTiVZr alloy film withstands 700 °C anneal. (AlMoNbSiTaTiVZr)50N50 films, 70 nm in thickness prevent the reaction up to 850 °C, which ranks second in all the published barrier materials. NbSiTaTiZr films, 20 nm in thickness is effective up to 800 °C. These results demonstrate that high-entropy alloy/nitride possess superior properties in this regard and have good resistance to the invasions from both sides, namely the silicidation reaction from one side and Cu in-diffusion from the other. High-entropy alloy/nitride have good resistance to silicidation because the atomic pairs in them have large enthalpy of mixing and the multi-element design brings in large entropy of mixing, which both lowers their free energy. As for the resistance against Cu in-diffusion, we suggest that it originates from the marked thermal stability and high packing density of its amorphous structure, which avoids the formation of crystalline phase and thus fast-diffusing grain boundaries. Therefore, high entropy alloy/nitride has excellent performances and thus this multi-element design concept may bring new insights to the design of future barrier materials.

Moreover, it was found that on many failed samples there exists many directional silicide line patterns. Such patterns were also reported by many authors, but none has tried to explain such phenomenon. We therefore proposed a model to describe the formation of these silicide patterns. The invasion of Cu begins from discontinuous points of the barrier layer. Cu then diffuses outwards in the Si single crystal from the break point. Due to the diffusion anisotropy in Si, Cu concentration develops faster along Si <110> and <100> and thus new silicides nucleates preferentially on these directions. Directional silicide line segments are therefore formed. When new silicides grow large enough they protrude from the substrate and break the barrier overlayer. Such barrier deformation and cracking offers new pathways for Cu, producing a new Cu source. In this way the growth of directional silicide line segments becomes self-sustaining. Moreover, the existence of directional silicide line segments indicates that the barrier failure is owing to extrinsic defects instead of the intrinsic disfunction of the barrier itself. Therefore, to obtain the intrinsic failure temperature of a barrier material one should pay extreme attention to the uniformity of the barrier film and the oxygen content in the annealing atmosphere.

並列關鍵字

high-entropy alloy ； diffusion barrier ； copper metallization ； multi-element nitrides ； high-entropy materials ； hard coating

參考文獻

3. Yamashita, K. and S. Odanaka. in Electron Devices Meeting, 1998. IEDM '98 Technical Digest., International. 1998.

4. Deutsch, A., H. Harrer, C.W. Surovic, G.A.H.G. Hellner, D.C.A.E.D.C. Edelstein, R.D.A.G.R.D. Goldblatt, G.A.A.B.G.A. Biery, N.A.A.G.N.A. Greco, D.M.A.F.D.M. Foster, E.A.C.E. Crabbe, L.T.A.S.L.T. Su, and P.W.A.C.P.W. Coteus. in Electron Devices Meeting, 1998. IEDM '98 Technical Digest., International. 1998.

6. Saraswat, K.C. and F. Mohammadi, Electron Devices, IEEE Transactions on, 1982. 29(4): p. 645-650.

7. Jeng, S.P., M.C. Chang, and R.H. Havemann. in MRS Symp. Proc. Adv. Metallization for Devices and Circuits. 1994.

11. Thomas, M.E., I.A. Saadat, and S. Sekigahama. in Electron Devices Meeting, 1990. IEDM '90. Technical Digest., International. 1990.