作者
Deepika Priyadarshini,S. Nguyen,H. Shobha,E. Todd Ryan,S. M. Gates,Huai Huang,James Chen,E. Liniger,S. Cohen,Chunhua Hu,Anita Madan,E. N. Adams,S. Molis,Thomas J Haigh,Griselda Bonilla,T. Standaert,D. Canaperi,A. Grill
摘要
Continuous shrinking of the interconnect dimensions with each technology node requires reduction in RC (resistance-capacitance) delays. Reduction in capacitance requirement at 45nm node was met by introducing a k 2.4 inter-layer dielectric (ILD) at 2X metallization level [1]. However, the material was not readily extendable to next technology node because of the need for even higher modulus at the advanced nodes. Thus, there was a need to develop a new pSiCOH k 2.4 ILD with improved mechanical properties and damage susceptibility for back end of the line (BEOL) capacitance reduction. IBM/Alliance team has developed a number of pSiCOH films using a variety of precursors. Generally, the porous ultra-low k films are prepared by using a subtractive approach [2] in which the film is deposited using a mixture of skeleton and porogen precursor and then cured to remove the labile organic fraction [2]. An alternative to the above approach is to use a single precursor molecule consisting of skeleton with embedded porogen [3]. Developing a robust pSiCOH film with k < 2.55 has always been a challenge with this approach. This paper reports the development of a pSiCOH k 2.4 film deposited using a single precursor showing superior time dependent dielectric breakdown (TDDB) performance with lower integrated k (lower capacitance) as compared to the k 2.55 ILD pSiCOH film at 80 nm, 56 nm and 48 nm pitch. Octamethylcyclotetrasiloxane (OMCTS) precursor used for fabricating dense SiCOH dielectrics (k= 3.0-2.7) was used to form low k=2.4 film. The modified PECVD deposition conditions along with optimized UV cure resulted in k 2.4 ILD film deposited using a single precursor henceforth referred to as OMCTS Ex k 2.4. The new OMCTS Ex k 2.4 film has only slightly lower modulus as compared to the k 2.7 film deposited using OMCTS and O2 while showing comparable plasma induced damage (PID). PID was measured by thickness change after HF wet etch removed the damaged layer caused by standard plasma. In general, good TDDB performance is obtained by lowering porosity, increasing %C and reducing PID [4]. Each of these properties has been carefully optimized for new OMCTS Ex k 2.4 film by tuning the deposition and the UV cure conditions. Fig. 1 compares the RC performance for ULK k 2.55, ULK A k 2.4 (another PECVD k2.4 ILD) and OMCTS Ex k 2.4 film at 10 nm node. ULK A k 2.4 shows higher PID as compared to k 2.55 resulting in higher capacitance. In contrast, the OMCTS Ex k 2.4 retains the capacitance benefit because of the lower PID for this film. Table 1 compares the extracted k value for k 2.55 and the k 2.4 ILD’s at 80, 56 and 48 nm pitch. OMCTS Ex continues to show lower extracted k number with no significant increase in value with technology node scaling. Fig. 2 shows the TDDB performance of these ILD’s at 10 nm node. The higher carbon content with low plasma induced damage resulted in dramatically better TDDB for OMCTS Ex than other films. Important to note that the TDDB performance of this film is almost 15X better than the TDDB performance of k 2.55 and ULK A k 2.4 ILD. Conclusion A new advanced single precursor OMCTS Ex k 2.4 pSiCOH ILD film has been developed to meet the integration scaling and reliability requirements. The new OMCTS Ex k 2.4 film shows better film properties than the reference k 2.55 and other 2.4 films and retains capacitance benefit by demonstrating an overall lower integrated k value as compared to the k 2.55 ILD. Acknowledgement This work was performed by the Research and Development Alliance Teams at various IBM Research and Development Facilities. References [1] S. Sankaran, et al , “A 45 nm CMOS node Cu/low-k/ultra low-k PECVD SiCOH (k=2.4) BEOL technology”, Electron Devices Meeting, IEDM International , 2006 [2] A. Grill and C. Patel, “Ultralow dielectric constant pSiCOH films prepared with tetramethylcyclotetrasiloxane as skeleton precursor,” J. Appl. Phys., vol. 104, pp. 024113-9, 2008 [3] S. Nguyen, S. Gates, D. Neumayer, and A. Grill, US patent 7,491,658, 2009 [4] E.G. Liniger, et al , “TDDB extendibility of ULK materials to 14nm BEOL and beyond,” Adv. Metal Conf. 2012 Figure 1