Study of Ru barrier failure in the Cu/Ru/Si system
- PDF / 412,029 Bytes
- 7 Pages / 585 x 783 pts Page_size
- 33 Downloads / 131 Views
T. Sritharan, S.G. Mhaisalkar, and E. Phoon School of Materials Science and Engineering, Nanyang Technological University, Singapore 639798
L. Chan Chartered Semiconductor Manufacturing Ltd., Woodlands Industrial Park D, Singapore 738406 (Received 11 February 2007; accepted 23 April 2007)
The reaction mechanisms and related microstructures in the Cu/Si, Ru/Si, and Cu/Ru/Si metallization system were studied experimentally. With the help of sheet resistance measurements, x-ray diffraction, field-emission scanning electron microscopy, secondary ion mass spectroscopy, and transmission electron microscopy, the metallization structure with Ru barrier layer was observed to fail completely at temperatures around 700 °C, regardless of the Ru thickness because of the formation of polycrystalline Ru2Si3 followed by Cu3Si protrusions.
I. INTRODUCTION
The International Technology Roadmap Semiconductor 2004 highlighted the importance of diffusion barrier layer development when they identified it as one of the key technology requirements for continuous scale down of integrated circuits. They predicted that a diffusion barrier thickness of 700
500 ⭈⭈⭈
⭈⭈⭈ 700
N/A ⭈⭈⭈
N/A 700
600
700
800
700
700
700
600
700
800
700
N/A
N/A
600
700
800
700
N/A
N/A
ated in the substrate from the native oxide layer and grows inward. This is the reason the Cu3Si phase was not detected on the surface by FE-SEM. This mechanism of barrier failure is schematically depicted in Fig. 12. Table I compares the failure temperatures obtained by the different characterization techniques. It is clear that resistivity measurement always indicates a lower failure temperature when compared to other microstructural methods. IV. CONCLUSION FIG. 12. Schematic depiction of the barrier failure mechanism in Cu/ Ru/Si: (a) Before barrier breakdown, (b) initiation of barrier breakdown with the nucleation of Ru2Si3, (c) complete consumption Ru to form Ru2Si3, and (d) diffusion of Cu through the Ru2Si3 to form Cu3Si in the substrate.
noted that an increase in Rs was measured at 600 °C, although no microstructure evidence for it is visible. After 700 °C annealing, neither the Cu layer nor the Cu/Ru interface was clear [Fig. 11(b)]. Cu diffusion appeared to have begun in earnest. The layer next to the substrate must consist of Ru2Si3, as demonstrated by XRD, but its thickness is not uniform. If one examines the Ru/Si interface closely, small crystals can be seen to grow into the substrate from the oxide layer [inset in Fig. 11(b)]. This could be the early formation of Cu3Si phase but may be too small to give XRD reflections. This further supports the notion that Cu diffusion begins before 700 °C and is enhanced when Ru2Si3 forms since this phase is not a continuous layer. The Cu3Si phase ripens at 900 °C, as shown in Fig. 11(c), indicating that the reaction continues to advance and the phase continues to grow into the Si substrate. Therefore, it is clear that good barrier properties could be derived from a continuous Ru layer for up to about 700 °C, an
Data Loading...
